Opening the Black Box: Internal Capital Markets and Managerial Power

Markus Glaser University of Konstanz

Universitätsstraße 10, PO Box 141 78457 Konstanz, Germany

Markus.Glaser@uni-konstanz.de

Florencio Lopez-de-Silanes EDHEC Business School

393 promenade des Anglais 06202 Nice, France

florencio.lopezdesilanes@edhec.edu

Zacharias Sautner University of Amsterdam &

Duisenberg School of Finance Roetersstraat 11

1018WB Amsterdam, the Netherlands z.sautner@uva.nl

This version: January 2011 ______________________ A previous version of this paper circulated under the title “Looking Inside a Conglomerate: Efficiency of Internal Capital Allocation and Managerial Power within a Firm.” We would like to thank Noël Amenc, Martin Artz, Marc Deloof (our EFA discussant), Holger Daske, Ingolf Dittmann, René Garcia, Michel Habib, Roman Inderst, Rajkamal Iyer, Li Jin, Abraham Lioui, Ernst Maug, Pierre Mella-Barral, Lionel Martellini, Enrico Perotti, Rafael Repullo, Markus Schmid, Henri Servaes, Amit Seru (our AFA discussant), Andrei Shleifer, Denis Sosyura (our FIRS discussant), Rene Stulz, Javier Suarez, Belen Villalonga, Alexander Wagner, Martin Weber, David Yermack, and seminar participants at the AFA Meetings, the FIRS Conference, University of Texas at Austin, University of Amsterdam, EDHEC Business School, Aalto University/Hanken School of Economics, London Business School, University of St. Gallen, Pfingsttagung der Hochschullehrer für Betriebswirtschaft Bremen, Tilburg University, University of Zurich, the Oxford Finance Summer Symposium, the European Summer Symposium in Financial Markets, the Annual Meetings of the European Finance Association, the Workshop on Internal Capital Markets at the University of Antwerp, the China International Conference of Finance, the Annual Meetings of the German Finance Association (DGF), the Workshop on Corporate Governance and Executive Compensation at the University of Mannheim, and the ECGTN Conference for helpful comments. Financial support from EDHEC Business School, the University of Amsterdam, and the Deutsche Forschungsgemeinschaft (DFG) is gratefully acknowledged (SFB 504 at the University of Mannheim). All errors remain our own.

Opening the Black Box: Internal Capital Markets and Managerial Power

Abstract

We analyze the internal capital markets of a multinational conglomerate to determine

whether more powerful unit managers enjoy larger allocations. We use a new dataset of

planned and actual allocations to business units to show that, although all unit managers

systematically over-budget capital expenditures, more powerful and better connected

managers obtain larger shares of cash windfalls and increase investment about 40% more

than their less powerful peers. Results survive robustness tests and are not explained by

differences in managerial abilities or an endogenous allocation of managers across units. Our

findings support bargaining-power theories and provide direct evidence of a source of capital

allocation frictions.

Keywords: Internal Capital Markets, Corporate Investment, Capital Budgeting, Managerial Power,

Agency, Influence Activities, Corporate Politics

JEL Classification Code: D80, G31, G34, L25

1

How do firms allocate resources across business units? Do units with better investment

opportunities receive larger capital allocations and invest more? Do units run by more

powerful or better connected managers enjoy bigger investment budgets? In this paper, we

seek to answer these questions with direct evidence from a newly constructed dataset, drawn

from the internal accounting system of a large multinational conglomerate, containing

information about planned and actual capital allocations to business units.

The modeling of internal capital markets has made substantial progress. Stein (2003)

and Maksimovic and Phillips (2007) provide comprehensive reviews of the bright and the

dark sides of internal capital markets, aspects initially studied by Gertner, Scharfstein, and

Stein (1994). The efficiency view of internal capital markets emphasizes the bright side. This

view holds that capital allocation is the result of pooling internally generated cash flows and

subsequently distributing funds optimally to units (Weston (1970), Williamson (1975),

Matsusaka and Nanda (2002), Maksimovic and Phillips (2002)). In these models, capital

allocation is determined mainly by the unit’s investment prospects. Through winner-picking

methods, internal capital markets add value, as the firm makes larger allocations to units with

greater investment opportunities (Stein (1997)).

A second view highlights the dark side of internal capital markets. The origin of this

idea can be traced back to Coase (1937) who suggested that power within a hierarchy may

impact internal firm policies. The proponents of this view argue that units run by more

powerful or better connected managers may get allocations larger than what is justified by the

investment opportunities they provide (Meyer, Milgrom, and Roberts (1992), Scharfstein and

Stein (2000), Rajan, Servaes, and Zingales (2000), Wulf (2009)). In these bargaining-power

models, unit CEOs prefer larger capital allocations (for rent-seeking or empire-building

2

reasons) and wield influence in an attempt to obtain more funds for their units. Influence

activities are costly and inefficient because resources are spent trying to affect allocations and

because the resulting investment decisions may not be optimal.1

Although there has been theoretical progress in the modeling of internal capital

markets, empirical evidence is scarce. The current situation is similar to that in the literature

on internal labor markets, a literature described by Baker and Holmstrom (1995, p. 259) as

being in a phase of “too many theories, too few facts” and in need of “studies of personnel

records, supplemented by interviews and institutional facts.” As in that literature, Schoar

(2002) and Maksimovic and Phillips (2002, 2007) also argue for the need to look within the

corporation to understand governance structures and their connection to resource allocation

and management decisions throughout the entity.2 In this paper, we take these

recommendations with an approach similar to that in Baker, Gibbs, and Holmstrom (1994a,

1994b), who use internal labor data from a large firm to test internal labor market theories.

We put together new data for a multinational conglomerate and analyze in detail its capital

allocation process and the means by which more powerful and better connected business-unit

managers use the process to obtain larger allocations after cash windfalls at headquarters.3

1 Some inefficient internal capital markets models focus on the firm CEO as the source of the agency problem. These models predict that CEOs try to allocate more capital to units from which they can obtain higher private benefits of control (Stulz (1990)), expect future favors, or help them get entrenched (Shleifer and Vishny (1989)). 2 The need to look within firms has also been recognized in the management accounting literature. Ittner and Larcker (2001, p. 396), for example, write that “greater use of detailed data from a single or small number of organizations is also recommended. [….] such studies may provide the only means for obtaining the quantity and level of data needed to answer many managerial accounting research questions.” 3 Our approach to looking within the corporation is also taken by Gopalan, Nanda, and Seru (2007), who examine the means by which internal capital markets in Indian business groups help support financially weaker firms, by Cremers, Huang, and Sautner (2010), who look at capital allocation data in a banking group, and by Seru (2010), who looks at R&D productivity in conglomerates.

3

Our analysis relies on two new databases. The first, drawn from the proprietary

records of the internal management accounting system of the conglomerate, contains five

years of data on planned (i.e., budgeted) and actual (i.e., realized) capital allocations for each

of the firm’s twenty business units, which operate under five divisions. In addition to the data

on capital expenditures, we collected data on planned and actual marketing and R&D

expenditures, as well as on assets, sales, sales growth, EBIT, cash flows, and personnel. To

our knowledge, this is the first such dataset used in the internal capital markets literature.

Most of the variables in our dataset are rarely available even at divisional level.4 An

additional advantage of our data is that it allows us to analyze individual business units. As in

most other conglomerates, the divisions of our firm lack operating activities themselves and

act mostly as organizational umbrellas for the underlying business units. Business units are

the real centers of economic activity in the typical conglomerate: they originate budget

requests, receive capital allocations, make investments, and initiate production.

We complement this data with a second dataset of measures of managerial power and

connections for the forty-three business-unit CEOs working at the conglomerate over the

five-year period of our analysis. We construct six indices of managerial power and

connections. An initial set of measures is based on management profile data from BoardEx

and ZoomInfo, data we cross-check and complement with information from annual reports,

social networking web pages, and a systematic web and newspaper search.5 These measures

4 Segment reporting rules such as US-GAAP (SFAS 131/ASC 280) and IFRS (IAS 14/IFRS 8) require firms to report divisional information only on assets, sales, liabilities, depreciation, income, and capital expenditures. In practice, firms do not publish data by business unit and do not provide information on their budgets or investment plans. See Villalonga (2004) for a discussion of additional potential problems of using divisional data to study internal capital markets. 5 BoardEx data has been recently used in Cohen, Frazzini, and Malloy (2008), whereas ZoomInfo data is used in Cohen, Frazzini, and Malloy (2009).

4

proxy for managers’ career at the firm, their social network at the workplace, and the

similarity of their personal profiles to that of the CEO.6 We also take a survey of business-

unit CEOs, a survey we designed ourselves, to put together a second set of measures of

managerial power and connection. The survey, which contains both self-constructed

questions and questions previously used in the management and sociology literature (Podolny

and Baron (1997), Forret and Dougherty (2001, 2004)), makes it possible to construct

measures capturing a unit manager’s networking activities and his or her connections to

executive management.

Before we look at the effects of managerial power on capital allocation, we analyze

the firm’s financial characteristics and its investment behavior. Throughout the period of

analysis, there is no indication that the firm is financially constrained or in need of cash to

finance investment opportunities. The firm also exhibits characteristics similar to those

documented for other conglomerates. Consistent with the numbers in Lang and Stulz (1994)

and Berger and Ofek (1995), the conglomerate discount of our firm averages 15% over the

sample period. The business units of the conglomerate also overinvest and show a sensitivity

of investment to Tobin’s Q lower than that of matched stand-alone firms, as posited by

Berger and Ofek (1995), Rajan, Servaes, and Zingales (2000), and Ozbas and Scharfstein

(2010). These results suggest that the conglomerate may be exposed to frictions in internal

capital allocation similar to those to which conglomerates analyzed in the aforementioned

papers are exposed.

6 Several of these measures are similar to those used in papers such as those by Berger, Ofek, and Yermack (1997), Ryan and Wiggins (2004), Cohen, Frazzini, and Malloy (2008), Fracassi (2008), Landier, Nair, and Wulf (2009), and Bandiera, Barankay, and Rasul (2009).

5

For a better understanding of the causes of frictions in internal capital allocation, we

start by analyzing the budget allocation process at our conglomerate. Bower (1970, p. 305)

describes capital allocation within firms as a political process and argues that “top

management must manage its influence on [such] political processes.” Similarly, Jensen

(2003, p. 380) contains vivid descriptions of how power and negotiations influence the

budgeting process. He writes that “the budgeting process frequently consumes six months of

management time in negotiations, planning, and target-setting.” Our access to internal firm

documents helps us open up this “black box” and provide a detailed picture of the stages of

the allocation process and the role played by unit managers. During the budgeting stage,

business-unit CEOs provide initial allocation plans negotiated through organized and

extensive interaction with senior management before they are approved. Although the general

unit budget is approved and investments below a threshold can be carried out, making large

expenditures still requires additional justification and approval during the execution stage.

But this final stage of the process is shorter and less formalized and it leaves more room for

discretion and favors.

The main contribution of our paper is that it documents empirically the role of

managerial power and connections in this process and their effects on actual capital

allocations. We provide some of the first empirical evidence on two main fronts. First, we

compare approved planned capital allocations (i.e., budgets) and actual allocations and show

that managers use the standardized budgeting process to build buffers into their budgets.

Business-unit managers systematically over-budget capital expenditures and end up investing

6

less than their approved allocation nearly 85% of the time. These empirically documented

findings corroborate previous research based on interviews and laboratory experiments.7

Second, we show that, although all unit managers try to use their excessive capital

budgets to justify additional spending, units run by more powerful and better connected

managers obtain higher actual capital allocations at times of financial slack in the firm. To

measure financial slack, we use the method of Blanchard, Lopez-de-Silanes, and Shleifer

(1994), who analyze firms’ behavior after unexpected cash windfalls. During our sample

period, the conglomerate experiences eight substantial cash windfalls resulting from the sale

by headquarters of equity stakes in other companies whose lines of business are largely

unrelated to those of the units.8 Although the business units did not consider the windfalls in

their previously submitted and approved planned capital budgets, some of the additional cash

is available, ex post, for their investments, thereby creating financial slack.

Actual investment rates across all units increase on average by about 40% in quarters

in which there are cash windfalls. But this increase is far from homogenous, and a large share

of the variation is accounted for by the power and the connections of the units’ CEOs. For six

different indices of power and connections, we find that more powerful unit managers get

significantly larger shares of the windfalls for their own units and increase their actual

investments on average 40% more than their peers do. These magnitudes are estimated after

controlling for proxies for investment opportunities and unobserved business-unit

heterogeneity. Results are robust to alternative specifications.

7 Dunk and Nouri (1998) provide a review of this literature. Studies using surveys include those by Onsi (1973), Merchant (1985), or Dunk (1993). More recently, Van der Stede (2000), for example, uses questionnaires to measure over-budgeting. Studies using laboratory experiments include Young (1985), Chow, Cooper, and Waller (1988), and Waller (1988) or Stevens (2002). 8 The windfalls are very large: they are equivalent, on average, to 53% of the conglomerate’s annual cash flow and 252% of its annual capital expenditures.

7

An advantage of focusing on the effect of power and connections on the distribution

of windfalls rather than on planned capital budgets is that it mitigates reverse causality

problems; after all, it could be that historically high planned capital allocations could

themselves be a source of intra-organizational managerial power. An additional advantage of

our method is that by looking at the difference between actual and planned investment, we

control for the investment opportunities of the units, opportunities that should be reflected in

the planned capital budgets. A final advantage of our data is that it allows us to address two

additional potential concerns. First, we show that the results are not driven by endogenous

allocation of powerful managers to the units with best investment opportunities. Second, we

provide evidence that our measures of power do not seem simply to proxy for managerial

ability.

An explanation consistent with our findings is that approval of investments is easier to

come by when the conglomerate is awash in cash and planned budgets are not yet fully spent.

Such approval is more likely for units run by more powerful and well-connected managers.

We provide evidence of one of the channels through which powerful managers obtain such

larger allocations. This evidence lends weight to bargaining-power theories, which posit that

capital allocations are based partly on power and connections.

Our results complement the work of Rajan, Servaes, and Zingales (2000), who show

that diverse firms, which are more susceptible to power struggles, allocate more resources to

units with few investment opportunities and that of Duchin and Sosyura (2010), who show

that the influence of division managers affects investment efficiency and firm value. It is also

related to that of Xuan (2009), who studies how job histories of CEOs affect capital

allocations to divisions. Our paper likewise complements that of Graham, Harvey, and Puri

8

(2010), who survey CEOs and CFOs in an attempt to elucidate the factors affecting the

distribution of capital across corporate divisions. Finally, we provide evidence corroborating

several studies that describe the ways in which political activity, lobbying, and persuasion

affect capital budgeting (Pfeffer and Salancik (1974), Covaleski and Dirsmith (1986), Hirst

and Baxter (1993)) and the ways in which middle managers try to convince their superiors,

on both formal and informal occasions, of their need for bigger budgets, and fight to wield

influence over allocation decisions (Lukka (1988), Lumijärvi (1991)).

The paper is organized as follows. Section I describes the conglomerate, its capital

budgeting process, and our planned and actual allocation data. This section also presents the

evidence for systematic over-budgeting of capital needs. Section II presents our evidence on

the effects of managerial power and connections on the allocation of cash windfalls. Section

III analyzes the robustness of the results and discusses alternative explanations. Section IV

concludes.

I. The Conglomerate and its Capital Allocation

A. Organizational Structure of the Conglomerate

Our firm is a major international conglomerate with business around the world. It has

production plants in more than fifteen countries on four continents and has more than

100,000 employees. The organizational structure of the firm takes the multidivisional M-

form9 and the conglomerate operates with a headquarters, five product divisions, and twenty

9 The M-form structure involves “the creation of semi-autonomous operating divisions (mainly profit centers) organized along product, brand, or geographic lines” (Williamson (1981, p. 1555)). The alternative is a

9

business units (see Figure 1).10 Headquarters coordinates central corporate functions,

especially corporate strategy, internal capital allocation, outside financing, M&A, investor

relations, and legal affairs. It also holds and manages equity stakes in corporations outside the

conglomerate. The executive board, chaired by the CEO, is located at headquarters and has

the ultimate responsibility for the firm.11

[FIGURE 1 HERE]

The five product divisions beneath headquarters have no separate operating activities

themselves and act as umbrellas under which the business units operate. The divisions cover

five distinct industry sectors and are run by division CEOs who have no direct operating

responsibility; instead, they coordinate the activities of the business units. As in most

conglomerates, the business units are the real centers of economic activity within the firm;

they are responsible for investment, production, and sales. All of the units do business

beyond the borders of the country where the firm is headquartered. The number of business

units in a division ranges from one to eight, as illustrated in Figure 1. The divisions and their

business units have no access to the external capital market and hence cannot raise debt or

equity. The division and business-unit managers are not part of the executive board.

centralized, functionally departmentalized or unitary (U-form) structure. The M-form was pioneered in large format by the US conglomerates Du Pont and Sloan. 10 The conglomerate has two other business units which are not included in our analysis. Neither unit can be compared with the twenty other units in the sample. One of these two units is a pure consolidation unit, which aggregates miscellaneous activities. This unit does not participate in the planning process and does not have a consolidated budget. The other unit is a service unit that does not make investments. As neither was a real business unit, neither had a business-unit CEO over the entire sample period. Annual sales and total assets of each of these two units amount to less than 1% of the aggregate sales and assets of the other twenty units combined. 11 The firm did not change its CEO during the sample period of our analysis.

10

B. Financial Characteristics of the Conglomerate

To ensure that the conglomerate is similar to other big conglomerates, we compare a

set of its financial variables with those of other large conglomerates. As comparison firms,

we look at all non-financial conglomerates in the Dow Jones 30 and the Euro Stoxx 50.12 We

do so also to show that our firm is unlikely to be financially constrained and in need of cash

to finance investment by its units. Panel A of Table I shows that our conglomerate’s

aggregate investment rate, calculated as capital expenditures over total assets, is similar to

those of the median conglomerate in the Dow Jones 30 and the lower-end conglomerates in

the Euro Stoxx 50. Apart from having relatively low leverage and high cash holdings, the

conglomerate also has a relatively high dividend payout ratio. Its dividend was constant for

the first three years of our sample and increased substantially in both 2005 and 2006.

Furthermore, the firm initiated a share repurchase program in 2005 and paid out nearly to

€1.4 billion to its shareholders. Taken together, these circumstances suggest that our firm is

unlikely to be financially constrained. As argued in Hovakimian and Titman (2006), these

considerations are important, as the cash windfalls, which we exploit in the next section,

could be considered not exogenous to the business units’ investments if the firm were

financially constrained.

[TABLE I HERE]

For a further comparison, we calculate the firm’s conglomerate discount. Using the

method of Berger and Ofek (1995), we find that the conglomerate discount is 15% over the

12 For this comparison, we consider all non-financial firms that operate at least in two different segments as conglomerates.

11

sample period.13 This figure is comparable to those documented for other large US and

European conglomerates (Berger and Ofek (1995), Lang and Stulz (1994), Hoechle et al.

(2010), Glaser and Müller (2010)). In fact, our conglomerate comes close to matching the 13-

15% discount found for the average US conglomerate in Berger and Ofek (1995).

In Panels B and C of Table I, we follow the classic papers in the internal capital

markers literature and show some of the potential causes of such a discount. Instead of using

divisional data, we use our more detailed business-unit information. Panel B compares the

annual investment rates of the business units and those of matched stand-alone firms

operating in the same industry as the units.14 We find that the business units of the

conglomerate invest more than comparable stand-alone firms. The evidence is in line with the

arguments in Jensen (1986) and Stulz (1990), who suggest that overinvestment in

conglomerates is a potential source of the value loss from diversification.

Panel C compares the sensitivity of the investments of business units and that of

matched stand-alone firms. We use a specification drawn from Ozbas and Scharfstein (2010)

and pool data for the business units and matched stand-alone firms. We then run regressions

of investment on investment opportunities (imputed Tobin’s Q), cash flow, a dummy that

equals one if an observation is from a unit of the conglomerate, and the interaction of the

13 To obtain this figure, we take the approach suggested by Berger and Ofek (1995) to compare our conglomerate and matched focused firms. For each of the five divisions, we collect data on all publicly-listed stand-alone firms doing business in the same three-digit SIC code. We then calculate the conglomerate’s excess value (EV) defined as EV=ln(MV/IV). MV, the firm’s market value, is defined as the market value of equity plus the book value of debt. IV, the firm’s hypothetical imputed value, is defined as the sum of the imputed hypothetical values of each division. The hypothetical division values are calculated as the division sales times the median value of the market value-to-sales ratio of a division’s matched publicly-listed stand-alone firms. 14 Matched stand-alone firms are firms that operate in the same industries as the business units based on four-digit NAICS codes. They are considered stand-alone if the first two (in robustness checks also the first three) digits of the first and secondary SIC codes are identical. We rely on SIC codes to identify focused firms as secondary NAICS codes are not available. The number of business units included in the analysis is less than twenty, as matched stand-alone firms are not available for all business units.

12

dummy and the first two variables. As Ozbas and Scharfstein (2010) find for divisions, we

find that business units exhibit lower sensitivity of investment to Tobin’s Q, which suggests

that their investment is generally less responsive to measures of investment opportunities

than is that of stand-alone firms.15

Taken together, the numbers in Table I suggest that the financial characteristics, the

diversification discount, and the general investment behavior of our conglomerate are similar

to those of other conglomerates. In the next section, we attempt to open the “black box” of

the conglomerate’s capital allocation process for a better understanding of how resources are

distributed and to identify potential sources of frictions in internal capital allocation.

C. Internal Capital Allocation: Process and Data

C.1 Internal Capital Allocation Process

To better understand the mechanics of capital allocation we asked our firm to grant us

access to internal company documents describing the internal budgeting, allocation, and

execution processes and to provide us detailed data on planned and actual allocations. We

were given access to information describing the details behind the budget creation and

allocation processes of the firm, especially on (1) the elaboration of budgets; (2) the time

lines of budgeting meetings; (3) the managers who participate in these meetings; and (4) the

15 We also carry out an analysis at division level following the method suggested by Shin and Stulz (1998), to further compare the conglomerate of our study with those of other studies that rely on divisional data. We thereby analyze the conglomerate’s internal capital market by regressing the investment of a division on its own cash flow, proxies for divisional investment opportunities, and the aggregate cash flow of the other divisions. We also use sales growth and imputed divisional Tobin’s Q as measures of investment opportunities. Our estimates are consistent with those documented for the average conglomerate in Shin and Stulz (1998). The estimates show that the firm operates an active internal capital market as a division’s investment is related to the cash flow available at other divisions. Moreover, the investment of a division depends much more on its own cash flow than on the cash flow of the firm’s other divisions, which suggests that the internal capital market does not fully insulate a division’s investment from its own operating cash flows.

13

procedures for approving investments. With these documents, and several interviews of

managers—unit CEOs, controllers from headquarters, and unit controllers—involved in the

budgeting process, we were able to put together the details of the full internal capital

allocation process of the conglomerate.

The general strategy of the firm is developed by the executive board and serves as the

foundation for capital allocation and other major corporate decisions. The internal capital

allocation process itself consists of two general stages: a budgeting (or planning) stage and a

realization (or execution) stage. The steps of these two stages can be visualized in Figure 2

and the exact details of the full process are described in Appendix A-3.

[FIGURE 2 HERE]

In the budgeting or planning stage (Panel A of Figure 2), the firm transforms the

general corporate strategy into planned capital allocations (budgets). The transformation is

completed in a two-phase process supported by the elaboration of two documents for each

business unit: (1) the strategic outlook, and (2) the annual capital allocation plan.

The goal of the first phase of budgeting is to develop a three-year strategic outlook for

each business unit through discussions about the long-term strategy of the unit. In January of

each year, business-unit CEOs begin identifying long-term growth and investment

opportunities for their units. These business plans are then presented by the business-unit

CEOs to the respective division CEOs. In February, negotiations take place over these

opportunities; preliminary decisions on the business plans are made by the division CEOs in

March. Each division CEO and his business-unit CEOs then present the three-year business

plans of their units, using standardized forms, to the executive board and negotiate revisions

and adjustments. The negotiations are coordinated, scheduled, and structured by corporate

14

control at headquarters, a centralized support service which also checks and analyzes the

plans. In April, after negotiation with the executive board and the unit managers, the

executive board makes a final decision on the figures in the strategic outlook. The strategic

outlook includes general targets for planned investments and required resource allocations for

all units. This document serves as the foundation for the internal capital allocation process.

As described, the process is highly institutionalized and structured; decisions are made for the

three-year period starting in January of the subsequent year.

The second phase of budgeting consists of the elaboration of a more detailed annual

capital allocation plan for each business unit. This plan is a one-year investment plan

containing detailed resource allocations. The annual plan of a unit is equivalent to a detailed

breakdown of the capital allocation in the strategic outlook setting more concrete investment

budgets.16 This plan also translates into a complete set of unit balance sheets and income

statements, which are subsequently aggregated and consolidated at division and at firm level.

Debate over the annual allocation plan gets underway in June; division CEOs draw on the

strategic outlook to prepare investment and allocation targets for the coming year. In July,

business-unit and division CEOs then negotiate these allocation plans and make any

necessary adjustments. Preliminary budget decisions are made in August. As in the strategic

outlook, again on standardized forms, the division and business-unit CEOs present the capital

allocation plan to the executive board, and a second round of in-depth joint negotiations and

revisions takes place in September and October. As in the strategic outlook, the negotiations

16 Appendix A-7 presents a time-line of the budgeting stage at our firm during the years of our analysis. The figure shows details of the calendar and overlap between the strategic outlook and the annual capital allocation phases of the budgeting stage. The figure also reports the years for which we have planned and actual accounting data available.

15

are coordinated, scheduled, and structured by corporate control. Finally, in October or

November, the executive board decides on the investment budgets for the coming year. The

divisions and business units then receive the approved plan and prepare the implementation

for the coming year.

It is during the realization or execution stage (Panel B of Figure 2) that concrete

investments are made and the corresponding cash outflows are processed. Decisions here are

made much more quickly than they are during budgeting. All investments (buying or

replacing small machines or IT, for example) below a specific investment threshold can be

made, quickly and easily, with simple procedures at the discretion of the business units. The

thresholds depend on the divisions and on their capital intensity varying from 0.7 to 2.3 % of

the mean total assets of the business units. Although the annual capital allocation plan is the

basis for investments by the business units, the firm requires additional approvals for the

execution of any project in the plan that involves investments in tangible or intangible fixed

assets that exceed the investment threshold. If the proposed investment exceeds the threshold,

the business unit must prepare, for approval by the division CEO and the executive board, an

investment memo that contains (1) a document to be signed by the executive board, (2) a

document describing the investment, and (3) a net present value calculation. The project

description should explain the reason for and the risks of the investment and a discussion of

alternatives. Each business unit must show that the investment will generate at least a positive

NPV. For this reason, the company has constructed an NPV calculator, a spreadsheet

provided by headquarters that all units must use to evaluate projects.17 Though important, the

17 As the discount rate for the NPV calculation, the firm uses a project-specific weighted average cost of capital (WACC), provided by headquarters. The input on interest rates, cost of debt, and country risk premiums used

16

NPV of a project is not the only relevant approval criterion. In fact, when it comes to deciding

which positive NPV projects are to be carried out, other more loosely defined criteria, such as

the strategic fit of a project, its cash-flow structure, or the past performance and reputation of

the business-unit CEOs behind the project, also play an important role. Taking into account

such a range of aspects to approve the execution of investments is not specific to our firm.

The recent CEO survey by Graham, Harvey, and Puri (2010), confirms that actual allocation

decisions in many firms are also based on cash-flow timing, market share, previous returns,

and even on manager reputation, manager confidence, and gut feel. After taking into account

all factors, the executive board makes a decision on the proposed investment and informs the

business unit of its final decision by automated email. The business unit may then go forward

with the investment by placing an order, signing a letter of intent or contract, or making a

payment. Headquarters allocates the necessary funds.

C.2 Internal Capital Allocation Data

The process described above provides the basis for a better understanding of our

capital allocation data. This data comes from the internal management accounting system and

covers quarterly data on planned (i.e., budgeted) allocations and monthly data on actual (i.e.,

realized) capital allocations for each of the twenty business units from 01/2002 to 12/2006.

Appendix A-7 reports the years for which we have planned and actual accounting data. Our

data on planned allocations is taken from electronic files containing the annual capital

allocation plans produced in the second phase of budgeting. Our data on realized allocations

for the calculation of the WACC is provided by the corporate treasury, at headquarters. Tax rates are provided by the corporate tax department, also at headquarters.

17

is taken from electronic files created by corporate control and is based on actual reporting by

the business units. Once aggregated, this data is used by the firm to put together its external

financial reporting.

[TABLE II HERE]

Panel A of Table II presents statistics for the data on planned allocations, whereas

Panel B shows statistics for data on realized allocations. Both panels show averages for the

business units in each of the five divisions. The exact definitions of all variables are

presented in Appendix A-1; all variables are calculated on quarterly observations. Since we

have twenty business units and twenty quarters, the maximum number of observations for

planned data is four hundred unit-quarters. Although we lack some data for a few business

units at the beginning of our sample period, the planned data on the main variables used in

the paper, including capital expenditures, sales, EBIT, and total assets is available for at least

359 business-unit-quarters. The coverage of data for planned R&D and marketing

expenditures, cash flows from operations, and personnel is not as wide, since these variables

were collected only as of 2004 and not at all for two business units.

To compare data for planned and actual allocations, the table includes only

observations that, for each variable, correspond to an exact match between the two forms of

data.18 Differences between the values for planned and actual allocations are shown in Panel

C. Values for planned allocations do not match those for realized allocations exactly; the

comparisons permit three particular observations. First, sales are overestimated at all business

units, and EBIT is overestimated for the units of three of the five divisions. Second, actual

18 Summary statistics across all unit-quarters for all available data are provided in Appendix A-4. Correlations are found in Appendix A-6.

18

sales growth and operational cash flows are not significantly different from planned values.

Finally, and, for our analysis, most importantly, planned capital expenditures are statistically

significantly higher than actual capital expenditures. This data suggests that business-unit

managers across the firm use the standardized capital allocation process systematically to

build slack into their budgets by over-budgeting for investments.19

To illustrate the systematic over-budgeting of capital expenditures, Figure 3 plots

annual averages of planned and actual capital expenditures for each of the twenty business

units. The graph shows that the units end up over-budgeting and investing below their

investment plans 83% of the time. The management accounting literature has called this

pattern “budgetary slack,” not to be confused with financial slack. Our findings, the first to

rely on hard data, corroborate those of previous research, which relies mostly on interviews

or laboratory experiments to show that managers try to over-budget for their investments

(Dunk and Nouri (1998), Onsi (1973), Merchant (1985), Dunk (1993), Young (1985), Chow,

Cooper, and Waller (1988), Waller (1988), Stevens (2002), Davila and Wouters (2005)).

[FIGURE 3 HERE]

The documented pattern of over-budgeting is arguably costly, as it may keep funds

available for all budgeted projects, which in turn may remain unused and be put into cash

accounts earning returns below the cost of capital (Antle and Eppen (1985)). Moreover, it is

problematic, as the budgeting process loses the “critical unbiased information that is required

to coordinate the activities of disparate parts of an organization” (Jensen (2003, p. 379)). The

19 We find no statisitically significant connection between over-budgeting and the power and connections of the business-unit CEOs, suggesting that building this slack is a conglomerate-wide pattern.

19

next section presents evidence that helps us better understand the behavior and incentives of

managers that may be behind this pattern of over-budgeting.

II. The Effect of Managerial Power on Capital Allocations

The pronounced and consistent difference between planned and actual capital

expenditures prompts questions about why unit CEOs would systematically ask for more

capital than they actually invest. To understand this behavior, one must connect the internal

capital allocation process and measures of managerial power. More specifically, we provide

evidence in this section that managers use their overblown capital budgets to justify

additional spending at times of financial slack. We show that more powerful and better

connected business-unit CEOs obtain substantially higher actual capital allocations for their

units when funds from cash windfalls are available.

A. Measuring Power and Connections in the Firm

There are forty-three business-unit CEOs working for the firm over the sample period.

Only 45% percent of the managers are still employed by the firm; the rest have left or retired.

Thirty-two of the managers worked as CEOs of only one business unit, nine as CEOs of two

units, and two as CEOs of three units. So there is a total of fifty-six different manager-

business-unit pairs for the forty-three CEOs and twenty business units.

To measure power and connections, we construct six indices capturing different

dimensions of the strength and internal network of the forty-three different business-unit

CEOs. Management profile data from BoardEx, ZoomInfo, social networking web pages

(LinkedIn, Facebook, MySpace), annual reports, and systematic web and newspaper searches

20

are used in the construction of the first three indices. These indices are available for at least

thirty-four of the forty-three unit managers. Despite a detailed search, the data for some

components of the indices was not available for all managers. Table III presents averages of

our six indices across unit managers, reported separately for each of the twenty business units

and by division. Summary statistics for each of the variables used in the indices are also

reported in Appendix A-2, and summary statistics for the indices across the full sample are

reported in Appendix A-4. All variables and indices are standardized or constructed in such a

way as to take values from zero to one.

[TABLE III HERE]

The first measure, Career at the Firm, proxies for a manager’s career at the firm and

for his or her social network at the workplace. It draws on the management and sociology

literature, which has underscored the importance of social networks at workplaces (Podolny

and Baron (1997)). It is formed by averaging three variables, each normalized to fall between

zero and one: (1) the number of months over which a manager coincided at a unit of the

conglomerate with a person who later became an executive board member; (2) the number of

years a manager has been working at the firm; and (3) the number of years a manager has

been working in a powerful position at the firm (division CEO, business-unit CEO, or head of

marketing). A measure of employment networks similar to the one we use for our first

variable is used by Fracassi (2008). For the second and third variables, we assume as in Ryan

and Wiggins (2004) or Berger, Ofek, and Yermack (1997) that the power and connections of

managers increase as their tenures lengthen and that this effect is particularly pronounced if

they are in powerful positions. We thereby attempt to capture the span of the more general

social networks of managers inside the company.

21

The second index, CEO Similarity, assesses the similarity of a manager’s profile to

that of the CEO of the firm (the firm had only one CEO during the sample period). It is

formed by averaging the following four dummy variables: (1) a dummy equal to one if a

manager speaks the native language of the CEO; (2) a dummy equal to one if a manager lives

in the country in which the CEO lives; (3) a dummy equal to one if a manager went to the

same university as the CEO; and (4) a dummy equal to one if a manager was a student in the

same academic discipline as the CEO. The first two variables of the index are inspired by the

work of Bandiera, Barankay, and Rasul (2009), who rely on workers’ and managers’

nationalities and the proximity of their dwelling places to assess the similarity of workers and

managers. We assume that managers are more likely to be closer to the CEO if they speak the

same language (which is not English) or live in close proximity to him (see Landier, Nair,

and Wulf (2009) for similar measures). The latter matters as the firm is a multinational

conglomerate with units around the world. The last two components measure personal

connections via shared educational networks and academic backgrounds in an attempt to

proxy, as in Cohen, Frazzini, and Malloy (2008), for the relationship of a manager and the

CEO. The third index, Power Index, is formed by averaging the previous two indices.

The three remaining indices are drawn from responses to a survey we took with the

support of the firm. The survey contained both questions we wrote and questions previously

used in the management and sociology literature (Podolny and Baron (1997), Forret and

Dougherty (2001, 2004)). In the summer of 2008, after several beta tests, we distributed a

questionnaire to all forty-three current and former business-unit CEOs. We guaranteed that

the survey answers would be analyzed with full confidentiality and anonymity for research

purposes only, and that they would not be traceable. In total, twenty business-unit CEOs—a

22

response rate of about 47%—returned the survey.20 The three survey-based indices are

available for at least sixteen managers. The survey questions used to construct the indices and

their summary statistics are reported in Appendix A-2 and summary statistics of the indices

across the full sample are reported in Appendix A-4.

Our Networker index measures the extent to which managers engage in networking

within the firm. The survey questions used to construct the index are similar to those in Forret

and Dougherty (2001, 2004). The index is formed by averaging the following five dummy

variables: (1) a dummy equal to one if a manager is a member of a fraternity; (2) a dummy

equal to one if a manager is a member of a social club inside the firm; (3) a dummy equal to

one if a manager regularly stops by the headquarters to say “Hello”; (4) a dummy equal to

one if a manager regularly accepts highly visible work assignments; and (5) a dummy equal

to one if a manager participated regularly in highly visible task forces or committees.

The fifth index, Division CEO Connection, measures how well a business-unit CEO is

connected to his or her division CEO. The index is formed by averaging the following

variables: (1) a dummy equal to one if a manager named the division CEO as a personal

mentor; (2) a dummy equal to one if a manager lunches at least occasionally with the division

CEO; and (3) a dummy equal to one if a manager meets the division CEO in person at least

every two weeks. The survey question underlying item (1) is based on a question used in

Podolny and Baron (1997), and the survey question underlying item (2) is based on a

question used in Forret and Dougherty (2001, 2004). Finally, the last index presented in

Table III is labeled Connected Networker and it is simply the result of averaging the previous

20As the firm supported our survey, we managed to get 65%—as opposed to only 22% of those who had left—of the unit CEOs who still worked at the firm to respond to our survey.

23

two indices. This index could be regarded as capturing an overall degree of the networking

abilities of a manager.

For all six indices Table III shows substantial variation both within and across

divisions. Some units seem to score consistently higher than others when it comes to the

power of their CEOs. It shows, for example, that the average values for the power and

connection measures of business units in divisions 1 and 2 are generally below the average

for the entire firm, whereas the values for those of divisions 4 and 5 are generally above it.

This pattern, however, is not uniform across all units in a division, as some business units in

divisions 1 and 2, for example, still score relatively high on some power measures. Moreover,

the table shows that there is considerable variation across the six measures within a unit,

indicating that they may be capturing complementary aspects of power and connection.

[TABLE IV HERE]

Panel A of Table IV provides an initial view of the role of managerial power and

connections. It shows that power and connections help explain differences in investment rates

and it serves as a point of departure for the more formal analysis in the next section of the

paper. The panel is constructed as follows. For each of the six indices of power or

connections, Panel A shows business units separated into two subsamples. First, for each year

and business unit we calculate the average value of each index. Second, we calculate the

median values across the business-unit-years and split the sample. For each measure of

power, a business-unit-year is in the high-power group if the power measure is above the

median of the sample and in the low-power group if it is below it. The panel shows average

investment rates for business units in the two subsamples and indicates whether the

difference in investment rates is statistically different from zero. For each subsample, we also

24

show the investment rates of matched stand-alone firms. Panel A provides evidence along

two dimensions. First, it suggests that units run by more powerful or better connected

managers invest significantly more than units run by less powerful or less well-connected

managers. Second, it suggests that units run by high-power managers also invest at

statistically significantly higher rates than matched stand-alone firms. By the criterion of the

index of Career at the Firm, for example, a powerful unit has an annual investment rate of

0.061, compared to only 0.033 for a unit with low power and compared to 0.400 of matched

stand-alone firms.

B. Managerial Power and Planned Capital Allocations

To understand intra-firm capital allocation, the management accounting literature

provides a different description and analysis of the budgeting (planning) and realization (or

execution) stage for capital allocation (Hirst and Baxter (1993), Marshuetz (1985) Mukherjee

(1988), Zaring (1996)). This literature also explains in depth how to design the budgeting

stage to diminish the detrimental effects of managerial power and connections. Most authors

agree that firms should try to set up a structured system of checks and balances to mitigate

managerial influence activities.21 Our conglomerate designed its budgeting stage, as

described in Section I, with characteristics resembling those suggested in the literature to

limit managerial influence activities during the planning phase.

21 Piercy (1987), for example, lists the following organizational tools that provide management mechanisms with which to improve the budgeting process: a formal organizational structure, the control of flows of information, avoidance of social interaction between people involved in the budget setting (between controllers at headquarters and unit managers), and a rewards and sanctions system associated with budget outcomes. Maccarone (1996) provides related suggestions. Similarly, textbooks in both management accounting and finance provide detailed suggestions on how to structure the budgeting process to reduce the effects of influence activities and corporate politics in capital allocation (Anthony and Govindarajan (2007), Brealey, Myers, and Allen (2006)).

25

[TABLE V HERE]

Table V presents regressions of ex ante planned investment on our measures of power

and connections, and a set of measures of investment opportunities.22 We run these

regressions to test more formally whether the mechanisms and structures used by our

conglomerate in the budgeting process reduce the effect of managerial influence on capital

allocation. The estimates show that planned investment is significantly related to measures of

future growth opportunities and profitability, indicating that the budgeting process attempts

to allocate funds to the units with the best prospects. These results are in line with

neoclassical investment models that suggest that corporate resources should go to the units

with the greatest growth opportunities (Maksimovic and Phillips (2002)). Interestingly, we

are unable to detect a statistical relationship between our measures of power and ex ante

capital allocations. These findings suggest that power and connections do not play a strong

role in the normal capital budgeting stage (that is, for data for planned allocations). It is thus

likely that the structured process described in Section I is able to neutralize managerial power

in the budgeting or planning phase.

Managerial power and connections may not affect planned allocations and the

budgeting stage, but one wonders if power is important in the realization (or execution) stage.

Although a structured budgeting phase may be designed to minimize the effects of corporate

politics on planned capital budgets, actual capital allocations in the realization phase may still

22 Having access to planning data allows us to use the profitability measures actually used by the firm to assess future opportunities. We control for lagged values of planned EBIT divided by total assets, a measure of planned capital intensity, planned sales growth, and the lagged deviation of planned from actual EBIT. This regression setup is consistent with management accounting research in which, for example, budget adjustments as a result of spending variances are analyzed (Lee and Plummer (2007)). We include a fourth-quarter dummy, as some of the business units sell large fractions of their products in the fourth quarter. The regressions include lagged values of the power and connection indices, as power in the current period may affect budgets drawn up for the following period.

26

be distorted by power and connections. Appendix A-6 presents some evidence of the

association between actual capital allocations and our power indices. The pair-wise

correlations of our six power indices and actual allocations are between 0.28 and 0.41. We

therefore need to take a closer look at the realization stage, a look that will also help us

understand why managers have incentives to over-budget investments.

C. Cash Windfalls and Capital Allocations

Though it seems difficult to deviate much from planned allocations and, in particular,

to exceed the plans, we argue and provide evidence that power may be particularly useful

when large windfalls arise and ad hoc decisions need to be made. In other words, although

the budgeting stage is formally structured to minimize the effects of power, the realization

stage may allow managers to exercise influence after cash windfalls resulting allocation.

The theoretical basis for this conjecture comes from the normative budgeting

literature. Based on an extensive literature review, Sisaye (1995) predicts when and how

power is most likely to matter. He writes that “when resources are scarce, the resource

allocation decisions give priority to technical and economic considerations” (p. 133). Capital

budgeting in firms is then based on a “rational choice model for resource allocation

decisions” (p. 133) and the allocations follow a formalized organizational process. By

contrast, he continues, “a political approach [is] commonly used in conflictual organizational

settings in which units bargain or compromise to receive a larger proportion of slack

organizational resources” (p. 151).23 Applied to our setting, Sisaye’s analysis suggests that

23 Slack resources, not be confused with budgetary slack, are resources in excess of those planned in the budgets.

27

resource allocation decisions after cash windfalls become a political process and may be

affected by the power and connections of business-unit managers. Zaring (1996, p. 233)

observes that “there is a lack of clarity about the way the models [used for capital budgeting]

should incorporate environmental consequences—the unexpected as well as the expected,” an

observation that suggests practices such as those in place at the firm we study are not out of

the ordinary.

In this section, we document empirically how managerial power and connections

affect capital allocations for investment after exogenous cash windfalls. During the sample

period, our conglomerate experienced eight large cash windfalls resulting from headquarters’

sales of equity stakes in other companies. The equity sales occurred in six different quarters.

Calculated across these six quarters, the average cash inflow from the windfalls comes to

€938 million. In all eight cases, the stakes were considered non-strategic holdings whose sale

allowed the firm to cut its exposure to those industries and exploit market opportunities. As

illustrated in Figure 4, the cash windfalls are substantial and are therefore likely to have an

impact on the operations of the conglomerate. Depending on the year, for example, the

windfalls represent between 24 and 84% of the firm’s total annual operational cash flow and

between 160 and 403% of its total annual capital expenditures. Appendix A-5 provides more

details of the windfalls.

[FIGURE 4 HERE]

The cash windfalls described provide a suitable scenario for testing the effects of

power and connections on capital allocation for four main reasons. First, the impact of power

on the distribution of windfalls rather than on planned capital budgets per se mitigates

potential reverse causality problems; after all, historically high planned capital allocations

28

could themselves be a source of intra-organizational managerial power. Second, cash

windfalls are largely exogenous to the divisions and their business units because the equity

stakes are not part of the unit's assets,24 they are held and managed by headquarters, and the

decision to sell them is made by corporate headquarters alone. Finally as Table I shows, our

firm is neither financially constrained nor in need of cash to finance investment opportunities.

The sale of the stakes is therefore not endogenous in the sense that they do not seem to be

sold to finance essential investment. Panel B of Appendix A-5 further dispel any notion that

the equity stakes are sold primarily to raise capital for the financing of projects. The panel

shows that only 19% of the money from the windfalls is used by the conglomerate for

internal investments. Twenty percent of the money is paid out to the firms’ shareholders in

the form of dividends and share repurchases, and another 20% is used to increase cash

holdings. 25

In Panel B of Table IV we provide a simple analysis linking the investment rates of

business units and the allocation of funds from cash windfalls. We split business units into

two subsamples based on the difference between capital expenditures in cash-windfall and

no-cash-windfall quarters. If the difference is large, it suggests that the capital expenditures

of the business unit increase greatly when proceeds from cash windfalls are available; if it is

small it suggests that they increase only little. A business unit belongs to the high-allocation

24 The lines of businesses of the companies in which the equity stakes are sold are unrelated to those of the business units of the conglomerate in five of the eight cases. In the remaining three cases, the only relationship to the business of the units is that the equity stakes sold belong to firms which were suppliers to two of the conglomerate’s units. Our results are robust to the exclusion of these two units from our analysis. 25 A related question is why the conglomerate's shareholders do not demand full payout of the cash windfalls. Part of the explanation may be the standard free cash-flow argument of Jensen (1986). Apart from having relatively low leverage, the firm is widely held and does not have block-holders owning above 5% of the shares. Additionally, the conglomerate operates in a country that, according to the investor protection indices by La Porta et al. (1998) and Djankov et al. (2008), is considered to have rather weak anti-director and anti-self-dealing rights.

29

subsample if the difference in allocation between cash-windfall and no-cash-windfall quarters

is above the median and it belongs to the low-allocation subsample if the difference is below

the median. Panel B shows that units where the difference in allocation from cash-windfall to

non-cash-windfall quarters is large have investment rates almost twice as high as those of

units that receive little from the windfalls; the investment rates are also substantially higher

than those of matched stand-alone firms. This simple analysis suggests that investment rates

may be connected to the distribution of cash windfalls.

D. Allocation of Cash Windfalls and Managerial Power

Figure 5 shows that cash windfalls seem to be associated with changes to actual unit

investment. Although actual investment is generally less than planned investment, actual

capital expenditure is substantially closer to the planned budgets during cash-windfall

quarters. This effect is statistically significant at the 1% level and uniform across all cash-

windfall quarters. If managerial power matters for the allocation of cash windfalls, more

powerful or better connected managers should receive a larger share of the windfalls. In other

words, the investments ultimately made by their units exceed planned investments by more

than the investments ultimately made by units run by managers who, as it were, are less likely

to spend an afternoon on the links with executive managers from headquarters or to wax

nostalgic with the CEO of the conglomerate for their long-gone days at the old alma mater.

[FIGURE 5 HERE]

To investigate this hypothesis formally, the regressions reported in Table VI use the

difference of actual minus planned capital expenditures over total assets as the dependent

variable. This measure is similar to the one in Rajan, Servaes and Zingales (2000) but instead

30

of using industry average expenditures, our analysis exploits the availability of business-unit

planned investment data. If planned allocations already reflect the investment opportunities of

the business units, our analysis has the advantage that it partially accounts for differences in

investment opportunities across units. All regressions in Table VI also account for

unobserved business-unit heterogeneity and year fixed effects. Regressions include a fourth-

quarter dummy, as some of the business units sell substantial fractions of their products

towards the end of the year.

[TABLE VI HERE]

The first three specifications in Table VI analyze the impact of of cash windfalls and

proxies of investment opportunities before introducing our measures of managerial power.

Consistent with the pattern presented in Figure 5, the regression estimates in column 1a show

that the cash windfalls move actual investments close to those planned.

The estimates in column 1b show that investment opportunities, as proxied by sales

growth and EBIT over assets, have a significant positive impact on how close to each other

actual and planned investments are. In particular, units with better opportunities are able to

make more of the investments budgeted for, whereas those with fewer opportunities seem not

to make as many budgeted investments. A one-standard-deviation increase in sales growth,

for example, implies an increase of 0.0015 (i.e., 15%) of actual minus planned capital

expenditures over assets. This change is equivalent to a 13% increase of the standard

deviation of our dependent variable. These results are in line with neoclassical investment

models (Maksimovic and Phillips (2002)), which suggest that corporate resources should go

to the units with the highest growth opportunities and that units with better opportunities

should therefore be closer to their plans. Specification 1b also shows that past deviations of

31

EBIT from its plan are also a significant determinant of the proximity of actual and planned

investments: business units that were farther away from their EBIT plans in the past are less

able to spend their entire investment budgets. This finding is consistent with the evidence in

the accounting literature, which suggests that the so-called profitability planning variances,

that is, differences between actual and planned EBIT, affect next-period actual capital

allocations (Anthony and Govindarajan (2007), Balakrishnan, Sivaramakrishnan, and

Sprinkle (2009)).

Specification 1c includes both the cash-windfall variable and investment opportunities

proxies. Estimates show that although the impact of sales growth weakens, while the

significance of EBIT and its past deviations from plan remain strong. Controlling for

investment opportunities reduces the statistical significance of cash windfalls to 10 percent,

but its magnitude remains virtually intact. A cash windfall of €938 million, which equals the

average cash windfall in a cash-windfall quarter, increases the difference of actual minus

planned investment over assets by 0.0041. Holding fixed planned investment, this estimate

suggests that actual investment increases by about this number.26 Relative to an average

quarterly business-unit investment rate of 0.01, calculated across all no-cash-windfall

quarters, our calculation says that the windfall leads to a substantial 41% increase in

investment (see Table VI Panel B).

The increase in capital expenditures in cash-windfall quarters is far from homogenous

from unit to unit; and some of the heterogeneity can be explained by different degrees of

power and connections of business-unit managers. Columns two to seven of Table VI include

26 We can assume a fixed planned investment rate as the cash windfalls were not incorporated in the annual capital allocation plan.

32

the cash-windfall variable, our power or connection indices, and interactions of the two. The

indices vary from regression to regression and are named in the columns above each

regression. The estimates show that, controlling for investment opportunities and business-

unit heterogeneity, managerial power and connections have a significant impact on the

allocation of cash windfalls. For all six indices we find that units run by more powerful or

well-connected managers receive a substantially larger share of the windfalls, as indicated by

the positive and significant coefficient of the interaction terms. Figure 6 complements this

evidence providing a set of partial scatter plots based on the regression results of Table VI.

These plots show that our results are not driven by outliers.

[FIGURE 6 HERE]

The estimated economic effects of power and connections on the cash-windfall

distribution are large. Suppose the firm experiments a windfall of €938 million in a quarter.

How do different degrees of managerial power or connection affect the difference between

actual and planned investments? To illustrate these economic effects, Panel B of Table VI

reports the changes in investment using the estimated coefficients of the regressions. Using,

for example, Career at the Firm, we find that the investment rate of a unit run by a CEO with

a power index at the twenty-fifth percentile of the distribution increases investment by only

0.0002 after a windfall. This change corresponds to an increase of 2% relative to the average

non-cash-windfall quarterly investment rate of 0.01. If, by contrast, a unit is run by a

powerful CEO with a power index at the seventy-fifth percentile, investment increases by

0.0039, which corresponds to a substantial 39% increase relative to the average non-cash-

windfall quarterly investment rate. The economic effects are similar and sometimes even

larger for the other five indices. These effects are also displayed in Figure 7. Across all

33

indices, more powerful and better connected unit managers increase their actual investments

on average 40% more than their peers do.

[FIGURE 7 HERE]

The estimates in Table VI provide strong evidence that well-connected and powerful

managers can capture substantial amounts from cash windfalls. For a better understanding of

why investments increase so substantially during cash-windfall quarters, we held several

structured interviews with managers and other important decision makers at the

conglomerate. One view expressed by our interviewees was that investment approval

procedures in the realization stage seem to be more lax in periods of large cash windfalls.

Some managers argued that the firm is thereby trying to avoid keeping large amounts of cash

from accumulating in its cash accounts, which could cause it to become a target for

shareholder activists. Our interlocutors also stated that the unspent money from regular

planned capital allocations is not available for expenditure by other business units. Proceeds

from cash windfalls, by contrast, are available to all units. If a unit does not use the available

funds for its investments, they are likely to be used by others.

Our results are consistent with these views. All managers have an incentive to over-

budget for their investments and to grab as large a share as possible of any windfalls. The

expectation of financial slack helps explain why managers have incentives to over-budget for

their investments: over-budgeting allows them to increase investments once cash windfalls

arise, usually without going over budget. The variety of investment criteria considered during

the execution stage provides flexibility and translates in a seemingly less stringent approval

34

process in the presence of cash windfalls. Importantly, our findings show that units run by

powerful managers succeed in obtaining more investment approvals during windfalls.27

III. Robustness and Alternative Explanations

A. Alternative Regression Specifications and Controls

In Table VII, we run a set of robustness checks of our main results. The table shows

only the coefficient and standard error of the power-windfall interaction term (Managerial

Power * Ln(1+Cash Windfall)). Specification A shows results using the same specification as

in Table VI but without control variables to ensure that the impact of our power measures is

not simply due to the specification used before. The following two specifications in Table VII

consider alternative controls to those used in Table VI for investment opportunities and the

importance of the unit in the firm, respectively. We exploit the availability of planned

investment data to account for the unit profitability expected by the firm. Specification B, for

example, uses planned sales growth as a proxy to control for the expected profitability of a

unit. The results are similar if we use planned EBIT over assets (not shown) instead. Results

are also robust to including cash flow from operations over assets and lagged values of EBIT

over assets. In specification C, we control for the (strategic) importance of a unit using its

size (total assets) as a proxy. Managerial power results are also robust to including other

measures of importance such as the size of unit sales or cash flows.

[TABLE VII HERE]

27 Since the conglomerate is headquartered in Europe, our results complement those in Graham, Harvey, and Puri (2010, p. 1), who find that corporate politics is more important for capital allocation in firms headquartered outside the United States.

35

The next three rows of the table present alternative econometric specifications to

those used in Table VI. Specification D reports results from pooled OLS regressions, without

business-unit fixed effects, but with standard errors clustered at business-unit level.

Specification E is estimated as in D but without clustering of standard errors to ensure that

our results are not affected by the relatively small number of clusters we use. In specification

F, we estimate an AR(1) process for the error term as capital expenditures are often very

persistent over time (Gatchev, Pulvino, and Tarhan (2010)) and to account for any possible

autocorrelation in the error term that may arise from the use of lagged control variables.

In the last two rows of the table, we address potential concerns about the design of our

analysis. Specification G uses actual capital expenditures over total assets as the dependent

variable. This specification suggest that units run by more powerful or connected managers

get higher capital allocations from the cash windfalls irrespective of their planned budgets.

Finally, specification H addresses the potential concern that three of the eight cash windfalls

may not be entirely unrelated to the lines of business of the units. In these three cases, the

windfalls result from the sale of equity stakes in firms that acted as suppliers to two of the

business units (see Appendix A-5). In specification H, we take a radical approach and

exclude all quarterly observations of these two business units. The estimates in Table VII

show that our results are robust to these alternative specifications.

B. Power as a Measure of Ability

We also run a series of tests to mitigate concerns that managerial power and

connection measures proxy for ability. We provide three pieces of evidence against this view.

First, if power and connection capture ability, our measures of power should have an effect

36

on planned capital budgets such that more able managers get larger allocations ex ante. The

findings in Table V show that managerial power is not connected to planned budgets.

Second, if the power variables capture ability, our proxies should be positively related

to the operating performance of the business units. To examine whether such a positive

relation exists, we regress the operating performance on our power proxies and a set of

controls in Table VIII. We take sales growth as a measure of operating performance.28 We

control for cash flow from operations, past investment, marketing expenditures, R&D

expenditures, total sales, and personnel intensity (number of employees relative to assets).

The estimates reported in Table VIII show that we cannot detect a positive and statistically

significant relation between performance and our measures of managerial power and

connections.

[TABLE VIII HERE]

Finally, we attempt to develop direct proxies for managers’ ability. One can argue that

more able managers are more likely to hold a degree from a top university and that higher

ability managers are more highly visible and more likely to be hired as non-executives by

other companies. To test for these two possibilities, we collect additional information on

whether unit CEOs hold a degree from an elite university and whether they hold non-

executive directorships outside the firm. Appendix A-1 provides the definition and method to

construct these two proxies. Summary statistics for the ability proxies are also reported in

Appendix A-4.

28 Our results are similar if we use EBIT/Total Assets as a proxy for performance. The advantage of using sales growth as the performance measure is that this variable is more easily verified and less prone to potential window dressing by the unit managers.

37

Although these two variables are certainly not perfect proxies, it is not unreasonable

to assume that they are correlated with ability. Some supporting evidence for this argument

comes from Appendix A-8, which shows that the two ability proxies are related to measures

of business-unit performance (i.e., sales growth) and planned capital allocation. These

estimates provide evidence that units run by more able managers perform better and receive

larger budget allocations

In Table IX, we add each of these two ability proxies as well as interaction terms

between the proxies and cash windfalls to the specifications presented in Table VI. Panel A

shows results when we use the elite-university variable and Panel B when we use the non-

executive-director variable. The estimates show that our results are robust to direct inclusions

of proxies for ability; some power variables even become more significant.

[TABLE IX HERE]

C. Endogenous Allocation of Managers to Business Units

Another alternative explanation for our findings may be the endogenous allocation of

powerful and well-connected managers to those business units with the best investment

opportunities. There are two main versions of this view. The first version, related to the

interpretation of power as ability, holds that headquarters allocates the most able managers to

the business units with the brightest prospects. In this scenario, ability drives the allocation of

managers. The second version argues that the most powerful managers lobby to be placed at

the units with the best investment prospects. In this scenario, the allocation of managers to

units is driven by their power rather than by their ability to run a specific unit.

38

Though plausible, our data suggests that neither of these two versions is likely to

explain our results. First, the evidence already presented in Tables V and VIII does not square

with either of these two versions. There does not seem to be a positive relationship between

(i) the power variables and planned capital budgets and between (ii) the power variables and

future performance. Second, the correlations presented in Appendix A-6 show that there is no

significant relationship between our indices of managerial power and actual sales, sales

growth, and EBIT/Assets. If anything, there seems to be a negative correlation between

managerial power and Tobin's Q. Finally, the internal job movements are rather rare in our

firm and the connections and power of managers who moved from one unit to another are not

significantly different from those of the rest of the managers. Of all forty-three business-unit

CEOs employed over the sample period, only eleven moved from one business unit to

running another. If we compare the indices of power and connection of the eleven managers

who moved from one unit to another and those of the managers who did not move, we detect

no significant between-group differences.

D. Capital Expenditures versus Marketing and R&D Expenditures

Our analysis has focused on the effects of power and connections on the allocation of

windfalls for capital expenditures. Our focus follows that of the previous literature as capital

expenditures are typically associated with empire-building by CEOs and unit managers

(Jensen (1986), Shleifer and Vishny (1989), Scharfstein and Stein (2000)). They are visible

investments, associated with status and prestige.

So far, we have not looked into allocations for research and development (R&D) or

marketing expenditures. The nature of R&D and marketing expenditures is different from

39

that of capital expenditures. R&D investments are usually riskier than capital expenditures

and are subject to large uncertainty. Moreover, there is usually a great lag between the

initiation of an R&D project and the point at which it begins to bear fruit. This lag may also

account in part for managers’ failure to invest more heavily in R&D (Hirshleifer and Thakor

(1992)). Marketing expenditures are also different from capital expenditures, additional

marketing may have positive externalities on other units of the firm, whereas the effects of

additional capital expenditures are felt mostly at the unit making the expenditures. For all of

these reasons, bargaining-power theories usually model capital expenditures to analyze the

relationship between power or connections and capital allocations (Scharfstein and Stein

(2000)).

We do not expect R&D and marketing expenditures to change significantly after the

arrival of windfalls. Evidence consistent with these expectations is provided in Appendix A-

9, which shows the cash-windfall regressions of Table VI with R&D (Panel A) and marketing

expenditures (Panel B) instead of capital expenditures. These two dependent variables are

also constructed as the difference between actual and planned expenditures over assets. The

estimates in column 1 of Panels A and B show that cash windfalls do not seem to affect the

deviation of actual from planned R&D and marketing expenditures. This lack of significance

is consistent with evidence suggesting that firms smooth R&D expenditures over time and

that shocks to cash flows do not seem to affect R&D expenditures (Lach and Schankerman

(1989), Hall (2002)). Moreover, R&D expenditures are usually conducted based on long-term

plans and often constitute a fixed fraction of sales (Anthony and Govindarajan (2007)).

Furthermore, R&D expenditures are very difficult to adjust as they mainly represent labor

40

expenses for scientists and engineers who are difficult to hire on short notice (Lach and

Schankerman (1989)). Similar arguments apply for marketing expenditures.

For all the reasons alluded above, we also do not expect power and connections

themselves to have an effect on the distribution of the cash windfalls for R&D or marketing

expenditures. Columns two to seven of both Panels of Appendix A-9 show evidence

consistent with this expectation. Two additional structural features present in many

conglomerates may also help explain the lack of impact of windfalls and managerial power

on R&D and marketing expenses. First, research activities are often shared by business units

in research-competence centers, so R&D expenditures, too, are less likely to be determined

by individual unit managers than are capital expenditures. Second, a large part of marketing

expenditures focus on promoting the brand name of the firm. So marketing expenses are

much less at the discretion of managers within the firm and less exposed to internal power

struggles.

IV. Conclusion

Our analysis tries to open up the black box of internal capital markets and test

predictions of theories linking managerial power and connections to capital allocations. To

test these predictions, we combine a unique dataset from the internal management accounting

system of a large international conglomerate, information on the backgrounds of

management, and internal survey data. We provide some of the first empirical evidence

showing that although all managers use the standardized budgeting process to build buffers

into their budgets, units run by more powerful and better connected managers are able to

obtain substantially higher capital allocations at times of firm financial slack.

41

Our empirical findings provide hard evidence that managerial power and connections

may be a major source of frictions in internal capital allocation. In its attempts to understand

the most important factors affecting the internal distribution of capital across corporate

divisions, our paper complements the recent CEO survey by Graham, Harvey, and Puri

(2010), and a large descriptive literature on the ways in which political activity, lobbying, and

persuasion affect capital budgeting (Pfeffer and Salancik (1974), Covaleski and Dirsmith

(1986), Hirst and Baxter (1993)).

Our findings open up an interesting new area for future research, since regular capital

allocation by which budgets are planned seems not to be significantly distorted by power. It

is possible that normal capital allocation already includes checks and balances whose aim is

to reduce managerial influence activities. In fact, our conglomerates’ capital allocation

process includes sophisticated and institutionalized procedures for the setting of ex-ante

planned allocations closely resembling those of textbook recommendations (Brealey, Myers,

and Allen (2006), Anthony and Govindarajan (2003)). In contrast, our results show that

bargaining power is highly relevant for ad hoc distribution of surprise windfalls. It is thus in

the wake of such windfalls, less likely to be incorporated into the firm’s institutionalized and

structured processes for allocating capital, that managerial power may be wielded to greatest

effect.

The results presented here have several implications. Most narrowly, and from a

firm’s perspective, they may justify policies of automatic job rotation, as described in Stein

(2003) for General Electric. Such policies usually mean that managers are automatically

moved from one division to another, thereby reducing their incentives to bargain for large

42

capital allocations for their current units. Another implication for managerial practice is to

formulate rules on how to proceed when unexpected cash windfalls arise.

More generally, our findings also have implications for the area of organizational

economics, and our understanding of the circumstances under which individuals may choose

to exercise power. They may contribute, for example, to an understanding of the effects of

lobbying faculty deans at universities when large donations are received, the success of

politically well-connected firms to obtain government funds when countries get IMF or

World Bank assistance (e.g., Faccio, Masulis, and McConnell (2006)), or the effects of

political power of interest groups or individuals on government spending during times of

unexpectedly high tax revenues.

43

References

Anthony, R. N., and V. Govindarajan, 2007, Management Control Systems (McGraw-Hill).

Antle, R., and G. Eppen, 1985, Capital rationing and organizational slack in capital

budgeting, Management Science 31, 163-174.

Baker, G., M. Gibbs, and B. Holmstrom, 1994a, The internal economics of the firm:

Evidence from personnel data, Quarterly Journal of Economics 109, 881-919.

Baker, G., M. Gibbs, and B. Holmstrom, 1994b, The wage policy of a firm, Quarterly

Journal of Economics 109, 921-955.

Baker, G., and B. Holmstrom, 1995, Internal labor markets: Too many theories, too few facts,

American Economic Review 85, 255-259.

Balakrishnan, R., K. Sivaramakrishnan, and G. Sprinkle, 2009, Managerial Accounting,

Wiley & Sons.

Bandiera, O., I. Barankay, and I. Rasul, 2009, Social connections and incentives in the

workplace: Evidence from personnel data, Econometrica 77, 1047-1094.

Berger, P. G., and E. Ofek, 1995, Diversification's effect on firm value, Journal of Financial

Economics 37, 39-65.

Berger, P. G., E. Ofek, and D. L. Yermack, 1997, Managerial entrenchment and capital

structure decisions, Journal of Finance 52, 1411-1438.

Blanchard, O. J., F. Lopez-de-Silanes, and A. Shleifer, 1994, What do firms do with cash

windfalls? Journal of Financial Economics 36, 337-360.

Bower, Joseph L, 1970, Managing the Resource Allocation Process (Harvard Business

School Press).

44

Brealey, R. A., S. C. Myers, and F. Allen, 2006, Principles of Corporate Finance (McGraw-

Hill, New York).

Chow, C., J. Cooper, and W. Waller, 1988, Participate budgeting: Effects of a truth-inducing

pay scheme and information asymmetry on slack and performance, The Accounting

Review 63, 111-122.

Coase, R., 1937, The nature of the firm, Economica 4, 386-405.

Cohen, L., A. Frazzini, and C. Malloy, 2008, The small world of investing: Board

connections and mutual fund returns, Journal of Political Economy 116, 951-979.

Cohen, L., A. Frazzini, and C. Malloy, 2009, Hiring cheerleaders: Board appointments of

“independent” directors, working paper, Harvard Business School.

Covaleski, M. A., and M. W. Dirsmith, 1986, The budgetary process of power and politics,

Accounting, Organizations and Society 11, 193-214.

Cremers, M., R. Huang, and Z. Sautner, 2010, Internal capital markets and corporate politics

in a banking group, Review of Financial Studies, forthcoming.

Davila, T., and M. Wouters, 2005, Managing budget emphasis through the explicit design of

conditional budgetary slack, Accounting, Organizations and Society 30, 587-608.

Djankov, S., R. La Porta, F. Lopez-de-Silanes, and A. Shleifer, 2008, The law and economics

of self-dealing, Journal of Financial Economics 88, 430-465.

Duchin, R., and D. Sosyura, 2010, Divisional managers and internal capital markets,

Working Paper, University of Michigan.

Dunk, A. S., 1993, The effect of budget emphasis and information asymmetry on the relation

between budgetary participation and slack, The Accounting Review 68, 400-410.

45

Dunk, A. S., and Nouri, H., 1998, Antecedents of budgetary slack: A literature review and

synthesis, Journal of Accounting Literature 17, 72-96.

Faccio, M., R. Masulis, and J. McConnell, 2006, Government bailouts and political

connections, Journal of Finance 61, 2597-2635.

Forret, M., and T. Dougherty, 2001, Correlates of networking behavior for managerial and

professional employees, Group Organization Management, 26, 283-311.

Forret, M., and T. Dougherty, 2004, Networking behaviors and career outcomes: Differences

for men and women? Journal of Organizational Behavior 25, 419-437.

Fracassi, C., 2008, Corporate finance policies and social networks, working paper, UT

Austin.

Gatchev, V., T. Pulvino, V. Tarhan, 2010, The interdependent and intertemporal nature of

financial decisions: An application to cash flow sensitivities, Journal of Finance 55, 725-

763.

Gertner, R., D. Scharfstein, and J. Stein, 1994, Internal and external capital markets,

Quarterly Journal of Economics 109, 1211-1230.

Glaser, M., and S. Müller, 2010, Is the diversification discount caused by the book value bias

of debt? Journal of Banking and Finance 34, 2307-2317.

Gopalan, R., V. Nanda, and A. Seru, 2007, Affiliated firms and financial support: Evidence

from Indian business groups, Journal of Financial Economics 86, 759-795.

Graham, J., C. Harvey, and M. Puri, 2010, Capital allocation and delegation of decision-

making authority within firms, Working Paper, Duke University.

Hall, B., 2002, The financing of research and development, Oxford Review of Economic

Policy 18, 35-51.

46

Hirshleifer, D., and A.V. Thakor, 1992, Managerial conservatism, project choice, and debt,

Review of Financial Studies 5, 437-470.

Hirst, M. K., Baxter, J. A., 1993, A capital budgeting case study: An analysis of a choice

process and roles of information, Behavioral Research in Accounting 5, 187-210.

Hoechle, D., Schmid, M. M., Walter, I. and D. Yermack, 2010, How much of the

diversification discount can be explained by poor corporate governance, Working Paper,

NYU Stern.

Hovakimian, G. and S. Titman, 2006, Corporate investment with financial constraints:

Sensitivity of investment to funds from voluntary asset sales, Journal of Money, Credit

and Banking 38, 357-374.

Ittner, C. D., and D. F. Larcker, 2001, Assessing empirical research in managerial

accounting: A value-based management perspective, Journal of Accounting and

Economics 32, 349-410.

Jensen, M., 1986, Agency costs of free cash flow, corporate finance, and takeovers, American

Economic Review Papers and Proceedings 76, 323-329.

Jensen, M., 2003, Paying people to lie: the truth about the budgeting process, European

Financial Management 9, 379-406.

Landier, A., V. Nair, and J. Wulf, 2009, Trade-offs in staying close: Corporate decision

making and geographic dispersion, Review of Financial Studies 22, 1119-1148.

Lang, L., and R. Stulz, 1994, Tobin’s Q, corporate diversification, and firm performance,

Journal of Political Economy 102, 1248-1280.

Lach, S., and M. Schankerman, 1989, Dynamics of R&D and investment in the scientific

sector, Journal of Political Economy 97, 880-904.

47

La Porta, R., F. Lopez-de-Silanes, A. Shleifer and R. Vishny, 1998, Law and finance, Journal

of Political Economy 106, 1113-1155.

Lee, T. M., and E. Plummer, 2007, Budget adjustments in response to spending variances:

Evidence of ratcheting of local government expenditures, Journal of Management

Accounting Research 19, 137-167.

Lumijärvi, O. P., 1991, Selling of capital investments to top management, Management

Accounting Research, 171-188.

Lukka, K., 1988, Budgetary biasing in organizations: Theoretical framework and empirical

evidence, Accounting, Organizations and Society 13, 281-301.

Maccarone, P., 1996, Organizing the capital budgeting process in large firms, Management

Decision 34, 43-56.

Maksimovic, V., and G. Phillips, 2002, Do conglomerate firms allocate resources

inefficiently across industries? Theory and Evidence, Journal of Finance 57, 721-767.

Maksimovic, V., and G. Phillips, 2007, Conglomerate firms and internal capital Markets, in

B. E. Eckbo, ed.: Handbook of Corporate Finance, 1: Empirical Corporate Finance

(Elsevier/North Holland)

Marshuetz, R. J., 1985, How Americans Can allocates capital, Harvard Business Review, 82-

91.

Matsusaka, J., and V. Nanda, 2002, Internal capital markets and corporate refocusing,

Journal of Financial Intermediation 11, 176–211.

Mukherjee, T.K., 1988, The capital budgeting process of large U.S. firms: An analysis of

capital budgeting manuals, Managerial Finance 14, 28-35.

48

Merchant, K., 1985, Budgeting and the propensity to create budgetary slack, Accounting,

Organization and Society 10, 201-210.

Meyer, M., P. Milgrom, and J. Roberts, 1992, Organizational prospects, influence costs, and

ownership changes, Journal of Economics and Management Strategy 1, 10-35.

Onsi, M., 1973, Factor analysis of behavioral variables affecting budgetary slack, The

Accounting Review 48, 535-548.

Ozbas, O., and D. Scharfstein, 2010, Evidence on the dark side of internal capital markets,

Review of Financial Studies 23, 581-599.

Pfeffer, J., and G.. Salancik, 1974, Organizational decision making as a political process: The

case of a university budget, Administrative Science Quarterly 19, 135-151.

Piercy, N., 1987, The marketing budgeting process: Marketing management implications,

Journal of Marketing 51, 45-59.

Podolny, J., and J. Baron, 1997, Resources and relationships: Social networks and mobility in

the workplace, American Sociological Review 62, 673-693

Rajan, R., H. Servaes, and L. Zingales, 2000, The cost of diversity: The diversification

discount and inefficient investment, Journal of Finance 55, 35-80.

Ryan, H. E., and R. A. Wiggins, 2004, Who is in whose pocket? Director compensation,

board independence, and barriers to effective monitoring, Journal of Financial Economics

73, 497- 524.

Scharfstein, D. S., and J. Stein, 2000, The dark side of internal capital markets: Divisional

rent-seeking and inefficient investment, Journal of Finance 55, 2537-2564.

Schoar, A, 2002, The effect of diversification on firm productivity, Journal of Finance 62,

2379-2403.

49

Seru, A., 2010, Firm boundaries matter: Evidence from conglomerates and R&D activity,

Journal of Financial Economics, forthcoming.

Shin, H.-H., and R. M. Stulz, 1998, Are internal capital markets efficient? Quarterly Journal

of Economics 113, 531-552.

Shleifer, A., and R. W. Vishny. 1989. Management entrenchment: The case of manager-

specific investments, Journal of Financial Economics 25, 123-139.

Sisaye, S., 1995, Control as an exchange process: A power control framework of

organizations, Behavioral Research in Accounting 7, 122-161.

Stein, J. C., 1997, Internal capital markets and the competition for corporate resources,

Journal of Finance 52, 111-133.

Stein, J., 2003, Agency, information and corporate investment, in G. Constantinides, M.

Harris, and R. M. Stulz, eds.: Handbook of the Economics of Finance (Elsevier).

Stevens, D., 2002, The effect of reputation and ethics on budgetary slack, Journal of

Management Accounting Research 14, 153-171.

Stulz, R., 1990, Managerial discretion and optimal financing policies, Journal of Financial

Economics 26, 3-27.

Van der Stede, W, 2000, The relationship between two consequences of budgetary controls:

Budgetary slack creation and managerial short-term orientation, Accounting,

Organizations and Society 25, 483-496.

Villalonga, B., 2004, Diversification discount or premium: New evidence from the business

information tracking series, Journal of Finance 59, 479-506.

Waller, W., 1988, Slack in participative budgeting: The joint effect of a truth-inducing pay

scheme and risk preferences, Accounting, Organization and Society 13, 87-98

50

51

Weston, F., 1970, Mergers and acquisitions in business planning, Rivista Internazionale di

Scienze Economiche e Commerciali 309-320.

Williamson, O., 1975, Markets and Hierarchies: Analysis and Antitrust Implications (Collier

Macmillan Publishers, New York).

Williamson, O., 1981, The modern corporation: Origins, evolution, attributes, Journal of

Economic Literature 19, 1537-1568.

Wulf, J. M., 2009, Influence and inefficiency in the internal capital market, Journal of

Economic Behavior and Organization 72, 305-321.

Xuan, Y., 2009, Empire-building or bridge-building? Evidence from new CEO’s internal

capital allocation decisions, Review of Financial Studies 22, 4919-4948.

Young, S., 1985, Participative budgeting: The effect of risk aversion and asymmetric

information on budgetary slack, Journal of Accounting Research 23, 829-842.

Zaring, O., 1996, Capital budgeting for the unexpected, Scandinavian Journal of

Management 12, 233-241.

Table I

Financial Characteristics of the Conglomerate and Investment of Its Business Units Panel A of this table shows the main financial ratios of the conglomerate and compares these ratios with those of non-financial conglomerates in the Euro Stoxx 50 and the Dow Jones 30 indices. Conglomerates are all firms from the indices with at least two divisions (segments) in different 2SIC codes. The reported ratios are the averages of annual data over the period 2002 to 2006. We also report the median, standard deviation, and the 25th and 75th percentiles for the conglomerates in each of the indices. Panel B presents statistics for annual investment rates (capital expenditures/total assets) of the conglomerate’s business units and of matched stand-alone firms over the period. Matched stand-alone firms are the firms included in Compustat (Global and North America) that, based on four-digit NAICS codes, operate in the same industries as the business units. Matched stand-alone firms are considered focused if the first two (2SIC) digits of the primary and secondary SIC codes are identical (results are very similar if we require that the first three (3SIC) digits are identical, but we have less matches). We rely on SIC codes to identify stand-alone firms as secondary NAICS codes are not available. The number of business units included in the analyses of Panel B is less than the twenty business units of the conglomerate (and the number of annual observations is thus less than one hundred), as matched stand-alone firms could not be identified for all years for three of the twenty business units. The last column of the panel presents the p-value of a difference-in-means t-test comparing the investment rates of the business units of our conglomerate and those of the matched stand-alone firms. Panel C follows Ozbas and Scharfstein (2010) and presents OLS regressions to analyze the sensitivity of investment rates of the business units of the conglomerate and those of matched stand-alone firms. Robust standard errors, clustered at firm level, are shown in parentheses. All variables in Panel C are Winsorized at 1%. Detailed definitions of each variable in the table are provided in Appendix A-1. *** indicates significance at 1%, ** indicates significance at 5%, and * indicates significance at 10%. Panel A: Comparison of Financial Ratios of Our Conglomerate and of Other Conglomerates Conglomerate Euro Stoxx 50 Conglomerates [N=32] Mean Median Std. dev. 25% 75% Capital Expenditures/Total assets 0.032 0.086 0.074 0.068 0.047 0.121 Total Debt/Total Assets 0.478 0.643 0.648 0.125 0.581 0.724 Cash Holdings/Total Assets 0.141 0.105 0.068 0.098 0.047 0.113 Dividends/EBIT 0.474 0.255 0.241 0.313 0.179 0.302 Conglomerate Dow Jones 30 Conglomerates [N=28] Mean Median Std. dev. 25% 75% Capital Expenditures/Total Assets 0.032 0.042 0.038 0.023 0.023 0.051 Total Debt/Total Assets 0.478 0.583 0.547 0.171 0.487 0.687 Cash Holdings/Total Assets 0.141 0.108 0.073 0.124 0.032 0.132 Dividends/EBIT 0.474 0.276 0.250 0.260 0.150 0.329 Panel B: Investment Rates of Business Units and Matched Stand-alone Firms Capital Expenditures/Total Assets Obs. Mean Median Std. dev p-value Business Units of the Conglomerate 88 0.046 0.031 0.040 Matched Stand-alone Firms 305 0.036 0.027 0.037

0.0171**

Panel C: Investment Sensitivity of Business Units and Matched Stand-alone Firms Capital Expenditures/Total Assets [1] [2] [3] [4] [2SIC] [3SIC] [2SIC] [3SIC] Imputed Tobin’s Q * Business Unit -0.0202** -0.0203** -0.0178* -0.0179* [0.0101] [0.0102] [0.0101] [0.0102] Imputed Tobin’s Q 0.0039 0.0040 0.0041 0.0042 [0.0025] [0.0026] [0.0026] [0.0027] Cash Flow from Operations/Total Assets * Business Unit -0.1036** -0.1038** -0.1029** -0.1050** [0.0502] [0.0510] [0.0509] [0.0517] Cash Flow from Operations/Total Assets 0.0232 0.0233 0.0234 0.0236 [0.0220] [0.0231] [0.0222] [0.0233] Constant 0.0325*** 0.0378*** 0.0341*** 0.0386*** [0.0075] [0.0086] [0.0073] [0.0084] Firm/Business-Unit Fixed Effects Yes Yes Yes Yes Year Fixed Effects No No Yes Yes Observations 352 260 352 260 Number of Firms & Business Units 80 66 80 66 Adjusted R-squared 0.145 0.155 0.141 0.150

Table II

Business-Unit Planned and Actual Accounting Data This table shows the planned (i.e., budgeted) and actual (i.e., realized) accounting data for the business units of the conglomerate. The data is obtained from the internal management accounting system of the conglomerate. The numbers reported in each column correspond to the means of all business units within each division between 01/2002 and 12/2006. Panel A reports planned accounting data, whereas Panel B reports actual accounting data. To ensure comparability, Panels A and B contain only the information of business units that have both planned and actual data. The number of observations for each variable and each division is reported in brackets underneath the respective value. Panel C reports the difference between planned and actual data and the significance levels of difference-in-means t-tests between planned and actual data. Detailed definitions of each variable are provided in Appendix A-1. *** indicates significance at 1%, ** indicates significance at 5%, and * indicates significance at 10%. Panel A: Plan Data Division 1 Division 2 Division 3 Division 4 Division 5 Planned Sales (millions) 795.84 218.21 1300.10 325.79 149.96 [60] [160] [19] [80] [72] Planned EBIT/Total Assets 0.010 0.027 0.018 0.066 0.090 [60] [160] [19] [80] [72] Planned Sales Growth 0.091 0.054 0.023 0.014 0.097 [57] [152] [18] [76] [68] Planned Cash Flow from Operations/Total Assets 0.012 0.036 0.066 0.073 0.128 [36] [96] [11] [48] [24] Planned Capital Expenditures (millions) 7.74 4.48 112.55 13.86 6.25 [36] [160] [19] [80] [64] Planned Capital Expenditures/Total Assets 0.011 0.008 0.029 0.021 0.027 [36] [160] [19] [80] [64] Planned R&D Expenditures/Total Assets 0.077 0.033 0.064 0.024 0.034 [36] [92] [11] [48] [24] Planned Marketing Expenditures/Total Assets 0.193 0.064 0.021 0.092 0.163 [36] [96] [11] [48] [24] Planned Personnel (# of employees) 4479 3810 33382 10554 2368 [36] [96] [11] [48] [24] Panel B: Actual Data Division 1 Division 2 Division 3 Division 4 Division 5 Sales (millions) 754.87 206.07 1132.43 314.39 138.39 [60] [160] [19] [80] [72] EBIT/Total Assets -0.020 0.014 0.006 0.061 0.089 [60] [160] [19] [80] [72] Sales Growth 0.042 0.056 -0.006 0.014 0.095 [57] [152] [18] [76] [68] Cash Flow from Operations/Total Assets -0.040 0.031 0.072 0.072 0.129 [36] [96] [11] [48] [24] Capital Expenditures (millions) 4.94 3.25 90.95 10.62 4.72 [36] [160] [19] [80] [64] Capital Expenditures/Total Assets 0.006 0.005 0.023 0.015 0.019 [36] [160] [19] [80] [64] R&D Expenditures/Total Assets 0.060 0.030 0.062 0.024 0.032 [36] [92] [11] [48] [24] Marketing Expenditures/Total Assets 0.156 0.062 0.021 0.092 0.150 [36] [96] [11] [48] [24] Personnel (# of employees) 4631 3709 34804 10745 2366 [36] [96] [11] [48] [24] Panel C: Differences (Actual– Plan) Division 1 Division 2 Division 3 Division 4 Division 5 Sales (millions) (Actual-Plan) -40.97*** -12.15*** -167.66*** -11.40*** -11.57*** [60] [160] [19] [80] [72] EBIT/Total Assets (Actual-Plan) -0.030** -0.012 -0.012* -0.005* -0.001 [60] [160] [19] [80] [72] Sales Growth (Actual-Plan) -0.050 0.002 -0.028 -0.001 -0.002 [57] [152] [18] [76] [68] Cash Flow from Operations/Total Assets (Actual-Plan) -0.052 -0.005 0.006 -0.001 0.001 [36] [96] [11] [48] [24] Capital Expenditures (millions) (Actual-Plan) -2.80*** -1.23*** -21.60* -3.23*** -1.53*** [36] [160] [19] [80] [64] Capital Expenditures/Total Assets (Actual-Plan) -0.005*** -0.003*** -0.005* -0.006*** -0.008*** [36] [160] [19] [80] [64] R&D Expenditures/Total Assets (Actual-Plan) -0.017** -0.003*** -0.002 0.000 -0.002** [36] [92] [11] [48] [24] Marketing Expenditures/Total Assets (Actual-Plan) -0.037** -0.003*** 0.000 0.000 -0.013** [36] [96] [11] [48] [24] Personnel (# of employees) (Actual-Plan) -152 101 -1422 -191 2 [36] [96] [11] [48] [24]

Table III

Indices of Managerial Power and Connections This table presents the indices of managerial power and connections for the business-unit CEOs of the conglomerate. The second and third columns of the table show the number of business units and the number of business-unit CEOs in each of the division of the conglomerate. There are twenty business units in the five divisions of the conglomerate. The total number of individual business-unit CEOs employed during our sample period is forty-three. The last six columns of the table report the mean values of the six indices of managerial power and connections for the managers in each business unit and each division of the conglomerate across the sample period from 01/2002 to 12/2006. Empty cells are missing data. The reports also in italics the p-values of Kruskal-Wallis tests, which test whether the indices are statistically different across business units within a division and across the business units of all divisions. Detailed definitions of each variable are provided in Appendix A-1. Managerial Power Indices

Division Business Unit

Number of business-unit

CEOs Career at the

Firm CEO

Similarity Power Index Networker Division CEO Connection

Connected Networker

1 1 1 0.497 0.500 0.498

1 2 2 0.264 0.250 0.257 0.200 0.333 0.267

1 3 5 0.144 0.413 0.278 0.200 0.333 0.267

Mean of Division 1 0.316 0.302 0.388 0.345 0.200 0.333 p-value (Kruskal-Wallis test)

0.0001

0.0001

0.0001

0.0001

0.0001

0.0001

2 1 3 0.318 0.594 0.456 1.000

2 2 5 0.292 0.708 0.500 0.433 0.250 0.200

2 3 2 0.081 0.250 0.165 0.400 0.667 0.533

2 4 4 0.116 0.125 0.120 0.400 0.333 0.367

2 5 2 0.045 0.438 0.241 0.450 0.333 0.392

2 6 4 0.095 0.344 0.220 0.500 0.583 0.542

2 7 2 0.108 0.463 0.285

2 8 2 0.103 0.500 0.318

Mean of Division 2 0.152 0.145 0.428 0.288 0.531 0.433 p-value (Kruskal-Wallis test)

0.0001

0.0001

0.0001

0.0001

0.0001

0.0001

3 1 3 0.265 0.500 0.382

Mean of Division 3 0.277 0.277 0.277 0.265 0.500 0.382 p-value (Kruskal-Wallis test)

n/a

n/a

n/a

n/a

n/a

n/a

4 1 3 0.640 0.800 0.720 0.429 0.714 0.571

4 2 2 0.213 0.750 0.482

4 3 2 0.892 0.575 0.946 0.570 0.950 0.760

4 4 4 0.280 0.563 0.421 0.378 0.444 0.411

Mean of Division 4 0.530 0.506 0.672 0.642 0.459 0.703 p-value (Kruskal-Wallis test)

0.0005

0.0005

0.0005

0.0005

0.0077

0.0005

5 1 2 0.569 0.500 0.535 0.600 0.667 0.633

5 2 2 0.738 0.500 0.619

5 3 3 0.357 0.463 0.410 0.600 0.765 0.682

5 4 3 0.051 0.488 0.269 0.442 0.474 0.458

Mean of Division 5 0.449 0.429 0.488 0.458 0.547 0.635 p-value (Kruskal-Wallis test)

0.0001

0.0001

0.0001

0.0001

0.0111

0.0001

Mean Across 5 Divisions 0.345 0.303 0.486 0.406 0.472 0.527

Median Across 5 Divisions 0.316 0.265 0.500 0.396 0.438 0.474

p-value (Kruskal-Wallis test) 0.0001 0.0001 0.0001 0.0001 0.0001 0.0001 Number of Business-unit CEOs with available data

35 35 36 34 17 16

Table IV

Business-Unit Investment Rates, Managerial Power and Cash Windfalls This table shows the average annual investment rates (capital expenditures/total assets) of the conglomerate’s business units (BU) and matched stand-alone firms (MSF). Average investment rates are calculated over the period from 2002 to 2006. Panel A provides an initial view of the role of managerial power and connections. The panel is constructed as follows. For each of the six indices of power or connections, Panel A splits business units into two subsamples. First, for each year and business unit we calculate the average value of each index. Second, we calculate the median values across the business-unit-years and split the sample. For each measure of power, a business-unit-year is in the high-power group if the power measure is above the median of the sample and in the low-power group if it is below it. The panel shows average investment rates for business units in the two subsamples and indicates whether the difference in investment rates is statistically different from zero. For each subsample, we also show the investment rates of matched stand-alone firms. Panel B splits business units into two subsamples based on the difference between capital expenditures in cash-windfall and no-cash-windfall quarters. A business unit belongs to the high-allocation subsample if the difference in allocation between cash-windfall and no-cash-windfall quarters is above the median and it belongs to the low-allocation subsample if the difference is below the median. Colum three of both Panels reports the difference in mean investment rates between the high- and low- groups and its statistical significance. Column six reports the difference in mean investment rates between the MSF of high- and low- BU groups and its statistical significance. Finally, columns seven and eight report whether the difference in investment rates of the business units in the high- and low- group and the investment rates of matched stand-alone firms (BU-MSF) is statistically different from zero, respectively. The numbers of observations in each group are reported in brackets. Detailed definitions of each variable are provided in Appendix A-1. *** indicates significance at 1%, ** indicates significance at 5%, and * indicates significance at 10%. Panel A: Power of Business Units and Investment Rates

Business units [BU] Matched Stand-alone Firms [MSF] Difference [BU-MSF] High

Power Low

Power Difference

[H-L] High

Power Low

Power Difference

[H-L] High

Power Low

Power

Career at the Firm

0.061 0.033 0.028*** 0.040 0.036 0.005 0.021*** -0.003

[43] [45] [170] [295]

CEO Similarity

0.057 0.037 0.020** 0.040 0.043 -0.003 0.017*** -0.007

[43] [45] [245] [215]

Power Index

0.050 0.043 0.007 0.040 0.036 0.004 0.010 0.007

[43] [45] [155] [290]

Networker

0.070 0.032 0.038*** 0.042 0.034 0.008 0.028*** -0.002

[29] [30] [155] [211]

Division CEO Connection

0.069 0.033 0.036*** 0.042 0.035 0.007 0.027*** -0.002

[29] [25] [155] [225]

Connected Networker

0.069 0.033 0.036*** 0.042 0.035 0.007 0.027*** -0.002

[29] [25] [155] [225] Panel B: Cash Windfalls and Investment Rates

Business Units [BU] Matched Stand-alone Firms [MSF] Difference [BU-MSF] High

Proceeds from Cash Windfalls

Low Proceeds

from Cash Windfalls

Difference [H-L]

High Proceeds

from Cash Windfalls

Low Proceeds

from Cash Windfalls

Difference [H-L]

High Proceeds

from Cash Windfalls

Low Proceeds

from Cash Windfalls

0.060 0.033 0.033*** 0.041 0.036 0.005 0.019*** -0.003

[43] [45] [160] [300]

Table V

Planned Investment and Managerial Power and Connection This table presents OLS regressions for the twenty business units of the conglomerate. The dependent variable in all regressions (shown in the first row) is a business-unit’s quarterly planned capital expenditures over its total assets over the period from 01/2002 to 12/2006. We report seven regressions. The first regression does not include any managerial power index as a control. The following six regressions successively control for the managerial power indices indicated in the second row: Career at the Firm; CEO Similarity; Power Index; Networker; Division CEO Connection; and Connected Networker. All regressions also control for: Planned EBIT/Assets (lag); Planned Personnel/Total Assets; Planed Sales growth; EBIT Deviation from Plan/Total Assets (lag); Fourth-Quarter Dummy; and a constant (not shown). The number of business-unit quarters and business units in each regression are reported at the bottom of the table. Robust standard errors, clustered at business-unit level, are shown in brackets. All variables are Winsorized at 1%. Detailed definitions of each variable are provided in Appendix A-1. *** indicates significance at 1%, ** indicates significance at 5%, and * indicates significance at 10%. Dependent Variable: Planned Capital Expenditures/Total Assets

Managerial Power Indices: None Career at the Firm

CEO Similarity

Power Index Networker

Division CEO

Connection Connected Networker

[1] [2] [3] [4] [5] [6] [7]

Managerial Power Index (lag) -0.0089 0.0028 0.0012 0.0010 0.0017 0.0055

[0.0066] [0.0044] [0.0055] [0.0022] [0.0024] [0.0045]

Planned EBIT/Total Assets (lag) 0.0320** 0.0238** 0.0328** 0.0255** 0.0383** 0.0333* 0.0330*

[0.0116] [0.0084] [0.0121] [0.0089] [0.0155] [0.0164] [0.0161]

Planned Personnel/Total Assets 0.0005** 0.0006*** 0.0005** 0.0006** 0.0007 0.0006 0.0006

[0.0002] [0.0002] [0.0002] [0.0002] [0.0004] [0.0004] [0.0004]

Planned Sales Growth -0.0001 -0.0025 -0.0003 -0.0039 0.0069 0.0043 0.0043

[0.0057] [0.0031] [0.0056] [0.0040] [0.0081] [0.0087] [0.0087]

EBIT Deviation from Plan/Total Assets (lag) 0.0069 0.0128 0.0086 0.0050 0.0199 0.0104 0.0059

[0.0306] [0.0236] [0.0299] [0.0264] [0.0472] [0.0569] [0.0567]

Fourth-Quarter Dummy -0.0021** -0.0025** -0.0020* -0.0024** -0.0012 -0.0010 -0.0010

[0.0010] [0.0010] [0.0010] [0.0010] [0.0012] [0.0013] [0.0013]

Business-Unit Fixed Effects Yes Yes Yes Yes Yes Yes Yes

Year Fixed Effects Yes Yes Yes Yes Yes Yes Yes

Observations 215 184 207 183 107 99 99

Number of Business Units 18 16 18 19 11 10 10

Adjusted R-squared 0.068 0.099 0.062 0.083 0.018 -0.007 -0.006

Table VI

Cash-Windfall Regressions: Effects of Managerial Power and Connection Panel A of this table shows OLS regressions for the twenty business units of the conglomerate. The dependent variable in all regressions (shown in the first row) is the quarterly difference between a business unit's actual and planned capital expenditures over its total assets over the period from 01/2002 to 12/2006. We report nine regressions. The first three regressions (i.e., 1a, 1b and 1c) do not include any managerial power index as a control. The following six regressions successively control for the managerial power indices indicated in the second row: Career at the Firm; CEO Similarity; Power Index; Networker; Division CEO Connection; and Connected Networker. The last six regressions also control for the interaction of each of the managerial power indices and Ln(1+Cash Windfall). The coefficient of the interaction term and its standard error are multiplied by 1,000. All regressions, except the second one, also control for Ln(1 + Cash Windfall). All regressions, except the first one (i.e., 1a) also control for: Sales Growth; EBIT/Total Assets; EBIT Deviation from Plan/Total Assets (lag); Imputed Tobin's Q (lag); Fourth-Quarter Dummy; and a constant (not shown). Robust standard errors, clustered at business-unit level, are shown in brackets. All dependent variables and controls are Winsorized at 1%. Detailed definitions of each variable are provided in Appendix A-1. *** indicates significance at 1%, ** indicates significance at 5%, and * indicates significance at 10%. Panel B reports the economic effects of the managerial power indices, based on the estimates in Panel A. The economic effects are calculated for a cash windfall of €938 million, which equals the average cash inflow in a cash-windfall quarter during our period of analysis. The change in the dependent variable is calculated both if the managerial power index is low (25th percentile) and if it is high (75th percentile The increase in the dependent variable is compared with the average quarterly business-unit investment rate, calculated across all no-cash-windfall quarters.

Panel A: OLS Regressions Dependent Variable: (Capital Expenditures - Planned Capital Expenditures)/Total Assets

Managerial Power Indices: None None None Career at the

Firm CEO

Similarity Power Index Networker Division CEO

Connection Connected Networker

[1a] [1b] [1c] [2] [3] [4] [5] [6] [7] Managerial Power * Ln(1+Cash Windfall) 1.7392** 0.7924** 1.7080** 3.0234* 2.2885* 3.6446** [0.7870] [0.3508] [0.6459] [1.5632] [1.0789] [1.4372] Managerial Power 0.0017 -0.0029 -0.0006 -0.0002 -0.0035 0.0007 [0.0056] [0.0049] [0.0066] [0.0043] [0.0060] [0.0116] Ln(1+Cash Windfall) 0.0006*** 0.0006* -0.0001 0.0002 -0.0003 -0.0008* -0.0008* -0.0013** [0.0002] [0.0003] [0.0002] [0.0004] [0.0003] [0.0004] [0.0004] [0.0004] Sales Growth 0.0056*** 0.0032 0.0035* 0.0026 0.0037* 0.0006 0.0029 0.0022 [0.0018] [0.0022] [0.0019] [0.0022] [0.0019] [0.0051] [0.0046] [0.0047] EBIT/Total Assets 0.0227* 0.0248** 0.0232* 0.0208* 0.0248* 0.0053 0.0095 0.0097 [0.0115] [0.0097] [0.0128] [0.0106] [0.0135] [0.0238] [0.0246] [0.0244] EBIT Deviation from Plan/Total Assets (lag) -0.0558*** -0.0565*** -0.0662*** -0.0614*** -0.0586*** -0.0644*** -0.0703*** -0.0732*** [0.0172] [0.0172] [0.0183] [0.0176] [0.0165] [0.0180] [0.0192] [0.0225] Imputed Tobin's Q (lag) 0.0097 0.0095 0.0083 0.0093 0.0088 -0.0000 0.0023 0.0026 [0.0069] [0.0069] [0.0079] [0.0079] [0.0084] [0.0079] [0.0078] [0.0078] Fourth-Quarter Dummy 0.0025 0.0012 0.0013 0.0017 0.0011 0.0035 0.0026 0.0027 [0.0018] [0.0015] [0.0014] [0.0016] [0.0015] [0.0033] [0.0033] [0.0033] Business-Unit Fixed Effects Yes Yes Yes Yes Yes Yes Yes Yes Yes Year Fixed Effects Yes Yes Yes Yes Yes Yes Yes Yes Yes Observations 359 299 299 246 270 242 141 133 133 Number of Business Units 20 20 20 18 20 18 13 12 12 Adjusted R-squared 0.034 0.111 0.128 0.100 0.106 0.089 0.076 0.064 0.075

Table VI (continued)

Panel B: Economic Effects of Managerial Power (for a Cash Windfall of 938m EUR)

Managerial Power Indices: None None None Career at the

Firm CEO

Similarity Power Index Networker Division CEO

Connection Connected Networker

[1a] [1b] [1c] [2] [3] [4] [5] [6] [7]

Low Power and Connection

(Variable at 25th percentile) n/a n/a n/a 0.076 0.250 0.264 0.400 0.333 0.367

Estimated Change in LHS variable 0.0041 n/a 0.0041 0.0002 0.0027 0.0010 0.0028 -0.0003 0.0002

Change in investment relative to average 41% n/a 41% 2% 27% 10% 28% -3% 2%

investment rate in no-cash-windfall quarters (=0.01)

High Power and Connection

(Variable at 75th percentile) n/a n/a n/a 0.381 0.500 0.485 0.600 0.667 0.633

Estimated Change in LHS variable 0.0041 n/a 0.0041 0.0039 0.0041 0.0036 0.0069 0.0050 0.0069

Change in investment relative to average 41% n/a 41% 39% 41% 36% 69% 50% 69%

investment rate in no-cash-windfall quarters (=0.01)

Difference Economic Effects n/a n/a n/a 37% 14% 26% 41% 53% 67%

Table VII

Cash-Windfall Regressions: Robustness Checks This table presents regressions for the twenty business units of the conglomerate. The dependent variable (shown in the first row) in all regressions, except those in row F, is a business-unit’s quarterly planned capital expenditures over its total assets over the period from 01/2002 to 12/2006. The table presents nine alternative specifications (one in each row) of the regressions in Table VI. Each cell in the table shows the coefficient and the standard error of the dependent variable on the interaction term of the managerial power index (shown in the second row) and Ln(1+Cash Windfall). The coefficient of the interaction term and its standard error are multiplied by 1,000. All regressions, except those in row A, also control for: the managerial power index shown in the second row; Ln(1 + Cash Windfall); Sales Growth; EBIT/Total Assets; EBIT Deviation from Plan/Total Assets (lag); Imputed Tobin's Q (lag); Fourth-Quarter Dummy; and a constant. Specifications in row A report results from specifications including as additional controls: the managerial power index shown in the second row; Ln(1 + Cash Windfall); and a constant. Specifications in row B include Planned Sales Growth as an additional control. Specifications in row C include Total Assets as an additional control. Specifications in row D report results from pooled OLS regressions, without business-unit fixed effects, but with standard errors clustered at business-unit level. Specifications in row E are estimated as those in row D but without clustering of standard errors. Specification in row F estimate an AR(1) process for the error term. Specifications in row G use business-unit quarterly actual capital expenditures over total assets as the dependent variable. Specifications in row H exclude all observations of the two business units whose business line is not entirely unrelated to three windfalls resulting from the sale of equity stakes in firms that acted as suppliers to these two units. All regressions use quarterly data from 01/2002 to 12/2006. All dependent variables and controls are Winsorized at 1%. Detailed definitions of each variable are provided in Appendix A-1. *** indicates significance at 1%, ** indicates significance at 5%, and * indicates significance at 10%.

Dependent Variable: (Capital Expenditures - Planned Capital Expenditures) /Total Assets Managerial Power Indices:

Career at the Firm

CEO Similarity Power Index Networker

Division CEO Connection

Connected Networker

[1] [2] [3] [4] [5] [6]

1.2082* 0.8158** 1.3260** 2.4768* 1.9525** 3.1726** A. Same as in Table VI but without controls

[0.6477] [0.3417] [0.5413] [1.2451] [0.8589] [1.1416]

1.7303** 0.8049** 1.7037** 3.0137* 2.3100* 3.6851** B. Controlling for planned measures of inv. opportunities (planned sales growth) [0.7993] [0.3400] [0.6581] [1.5889] [1.1204] [1.5080]

1.6890** 0.7954** 1.6744** 2.9311* 2.2743* 3.5825** C. Controlling for size of BUs (total assets)

[0.7759] [0.3433] [0.6332] [1.5918] [1.0539] [1.4809]

1.3712* 0.9762*** 1.5908** 2.9814* 2.0183* 3.1408** D. OLS without fixed effects, standard errors clustered at BUs [0.6936] [0.3013] [0.5751] [1.3911] [0.9550] [1.3823]

1.3712* 0.9762* 1.5908** 2.9814*** 2.0183* 3.1408* E. OLS without fixed effects, no clustering of standard errors [0.8164] [0.5480] [0.7846] [1.1130] [1.1889] [1.6728]

F. AR(1) process for error term 2.1724** 1.2151 2.3452** 4.3305** 2.5124* 4.0659**

[0.8435] [0.9232] [1.0185] [1.9601] [1.2879] [1.8086]

1.2669** 1.3238** 1.6214** 3.5758 3.9046** 5.6325** G. Capital Expenditures/Total Assets as dependent variable [0.5690] [0.5625] [0.6472] [2.0128] [1.7432] [2.2646]

1.9323* 0.7983* 1.7535** 3.0234* 2.2885* 3.6482** H. Excluding all observations of BUs related

to the origin of the cash windfalls [0.9824] [0.3802] [0.7503] [1.5632] [1.0789] [1.4328]

Table VIII

Sales Growth and Managerial Power and Connection This table shows OLS regressions for the twenty business units of the conglomerate. The dependent variable in all regressions (shown in the first row) is a business unit’s quarterly sales growth over the period from 01/2002 to 12/2006. We report seven regressions. The first regression does not include any managerial power index as a control. The following six regressions successively control for the managerial power indices (lag) indicated in the second row: Career at the Firm; CEO Similarity; Power Index; Networker; Division CEO Connection; and Connected Networker. All regressions also control for: Cash Flow from Operations/Total Assets; Capital Expenditure/Total Assets (lag); R&D Expenditure/Total Assets; Marketing Expenditure/Total Assets; Sales; Personnel/Total Assets; Fourth-Quarter Dummy; and a constant (not shown). The number of business-unit quarters and business units in each regression are reported at the bottom of the table. Robust standard errors, clustered at business-unit level, are shown in brackets. All dependent variables and controls are Winsorized at 1%. Detailed definitions of each variable are provided in Appendix A-1. *** indicates significance at 1%, ** indicates significance at 5%, and * indicates significance at 10%. Dependent Variable: Sales Growth

Managerial Power Indices: None Career at the Firm

CEO Similarity

Power Index Networker

Division CEO

Connection Connected Networker

[1] [2] [3] [4] [5] [6] [7]

Managerial Power (lag) -0.0348 -0.0211 -0.0415 -0.1591** -0.0832 -0.1890

[0.0480] [0.0471] [0.0604] [0.0611] [0.0836] [0.1229]

Cash Flow from Operations/Total Assets 0.3582** 0.3898** 0.3621** 0.3604* 0.2073 0.2422 0.2226

[0.1576] [0.1816] [0.1671] [0.1730] [0.3526] [0.3424] [0.3425]

Capital Expenditure/Total Assets (lag) -4.5018*** -4.6282*** -4.1294** -4.4621*** -3.7297* -3.6421* -3.6154*

[1.3886] [1.4195] [1.4754] [1.3204] [1.7338] [1.8023] [1.7683]

R&D Expenditures/Total Assets -2.0536 -1.3396 -1.6973 -0.8320 -1.2708 -1.2213 -1.1423

[1.7338] [1.4911] [1.5921] [1.3781] [1.4747] [1.5162] [1.5556]

Marketing Expenditures/Total Assets 2.5874*** 2.2650*** 2.7482*** 2.4884*** 3.8092*** 3.7295*** 3.7570***

[0.8272] [0.7835] [0.7797] [0.7366] [1.0624] [1.0699] [1.0406]

Sales 0.0007*** 0.0007** 0.0006*** 0.0007** 0.0012** 0.0012*** 0.0012***

[0.0002] [0.0003] [0.0002] [0.0003] [0.0004] [0.0004] [0.0004]

Personnel/Total Assets -0.0068 -0.0058 -0.0170 -0.0188 -0.0512*** -0.0461** -0.0499**

[0.0077] [0.0092] [0.0111] [0.0134] [0.0160] [0.0168] [0.0176]

Fourth-Quarter Dummy 0.2080*** 0.1948*** 0.2104*** 0.1983*** 0.2088** 0.2050** 0.2073**

[0.0482] [0.0542] [0.0484] [0.0528] [0.0715] [0.0746] [0.0735]

Business-Unit Fixed Effects Yes Yes Yes Yes Yes Yes Yes

Year Fixed Effects Yes Yes Yes Yes Yes Yes Yes

Observations 390 304 341 296 175 167 167

Number of Business Units 20 19 19 19 14 13 13

Adjusted R-squared 0.485 0.491 0.499 0.505 0.555 0.551 0.554

Table IX Cash Windfall-Regressions: Controlling for Managerial Ability

This table presents OLS regressions for the twenty business units of the conglomerate. The dependent variable in all regressions (shown in the first row) is the quarterly difference between a business unit's actual and planned capital expenditures over its total assets over the period from 01/2002 to 12/2006. We report seven regressions. The first regression of each panel does not include any managerial power index as a control. The following six regressions of each panel successively control for the managerial power indices indicated in the second row: Career at the Firm; CEO Similarity; Power Index; Networker; Division CEO Connection; and Connected Networker. The last six regressions also control for the interaction of each of the managerial power indices and Ln(1+Cash Windfall). The coefficient of the interaction term and its standard error are multiplied by 1,000. All regressions also control for (not shown): Ln(1 + Cash Windfall); Sales Growth; EBIT/Total Assets; EBIT Deviation from Plan/Total Assets (lag); Imputed Tobin's Q (lag); and a Fourth-Quarter Dummy. Regressions in Panel A also include as controls an Elite University dummy and the interaction of Elite University and Ln(1+Cash Windfall). The coefficient of the interaction term and its standard error are multiplied by 1,000. , is equal to one if a manager holds a degree from an elite university. In Regressions in Panel B also include as controls an External Director dummy and the interaction term of Executive Director and Ln(1+Cash Windfall). The coefficient of the interaction term and its standard error are multiplied by 1,000. Robust standard errors, clustered at business-unit level, are reported in brackets. All regressions include a constants, but the coefficient of the constant is not shown. All dependent variables and the controls are Winsorized at 1%. Detailed definitions of each variable are provided in Appendix A-1. *** indicates significance at 1%, ** indicates significance at 5%, and * indicates significance at 10%.

Panel A: Elite University as Ability Control Dependent Variable: (Capital Expenditures - Planned Capital Expenditures) /Total Assets

Managerial Power Indices: None Career at the Firm

CEO Similarity Power Index Networker

Division CEO

Connection Connected Networker

[1] [2] [3] [4] [5] [6] [7]

Managerial Power * Ln(1+Cash Windfall) 1.7972** 0.7821 1.7998** 3.4190** 2.2065* 3.7861** [0.6968] [0.6058] [0.6508] [1.2376] [1.0066] [1.2954] Managerial Power 0.0005 -0.0041 -0.0026 -0.0092 -0.0030 -0.0041 [0.0037] [0.0033] [0.0044] [0.0056] [0.0053] [0.0096] Elite University * Ln(1+Cash Windfall) -0.0008 -0.0004 -0.0008 -0.0004 -0.0001 -0.0004 -0.0001 [0.0004] [0.0004] [0.0005] [0.0004] [0.0005] [0.0005] [0.0004] Elite University -0.0034 -0.0046** -0.0044* -0.0054** -0.0095** -0.0054* -0.0073*** [0.0021] [0.0020] [0.0024] [0.0023] [0.0036] [0.0030] [0.0016] Business-Unit Fixed Effects Yes Yes Yes Yes Yes Yes Yes

Year Fixed Effects Yes Yes Yes Yes Yes Yes Yes

Observations 274 246 270 242 141 133 133 Number of Business Units 20 18 20 18 13 12 12 Adjusted R-squared 0.132 0.117 0.130 0.107 0.069 0.057 0.068

Panel B: External Director Positions as Ability Control Dependent Variable: (Capital Expenditures - Planned Capital Expenditures) /Total Assets

Managerial Power Indices: None Career at the Firm

CEO Similarity

Power Index Networker

Division CEO

Connection Connected Networker

[1] [2] [3] [4] [5] [6] [7] Managerial Power * Ln(1+Cash Windfall) 1.7255** 0.9598* 1.6955** 3.3084** 2.2225** 3.7853** [0.7397] [0.5008] [0.6561] [1.3168] [0.9773] [1.2727] Managerial Power 0.0009 -0.0050 -0.0020 -0.0069 -0.0043 -0.0066 [0.0048] [0.0045] [0.0059] [0.0044] [0.0046] [0.0092] External Director * Ln(1+Cash Windfall) -0.0006 -0.0002 -0.0007 -0.0003 -0.0002 -0.0004 -0.0002 [0.0005] [0.0004] [0.0005] [0.0004] [0.0006] [0.0005] [0.0004] External Director -0.0026 -0.0034 -0.0030 -0.0036 -0.0048 -0.0030 -0.0045 [0.0023] [0.0024] [0.0028] [0.0028] [0.0032] [0.0039] [0.0038] Business-Unit Fixed Effects Yes Yes Yes Yes Yes Yes Yes Year Fixed Effects Yes Yes Yes Yes Yes Yes Yes Observations 295 246 270 242 141 133 133 Number of Business Units 20 18 20 18 13 12 12 Adjusted R-squared 0.139 0.107 0.121 0.096 0.066 0.055 0.065

Headquarters

Division 1

BU1

DivisionCEO

BUCEO

ExecutiveBoard

Division 2

DivisionCEO

Division 3

DivisionCEO

Division 4

DivisionCEO

Division 5

DivisionCEO

BU3

BUCEO

…

# 3

BU1

BUCEO

BU8

BUCEO

…

# 8

BU1

BUCEO

BU4

BUCEO

…

# 4

BU1

BUCEO

BU4

BUCEO

…

# 4

BU1

BUCEO

# 1

BU = Business Unit

CEO

Figure 1. Organizational Structure of the Conglomerate. This figure summarizes the organizational structure of the conglomerate. It shows the five divisions and the business units operating within the divisions. It illustrates, for example, that division 2 operates a total of eight business units. The divisions, which have no operating activities themselves, are run by division CEOs. The business units are the operating units and they are run by business-unit CEOs (BU CEOs). There are twenty business units in total.

Panel A: Budgeting or Planning Stage

Involved Party:

Involved Party:

March/April: Negotiations over long-term business plans of

Business Units

September/October: Negotiations over one-year investment and allocation targets of

Business Units

Executive Board (Headquarters)

October/November: Final decision on one-

year annual capital allocation plan for all

Business Units

June: Based on Strategic Outlook, prepare one-year concrete investment and

allocation targets

Objective: Three-year long-term strategic plan ("guide")

Objective: One-year detailed capital allocation plan

April: Final decision on long-term business

plans for Business Units (three-year strategic

outlook)

February: Negotiations over long-term

investment opport. Business Units March: Prelim.

decisions on long-term business plans of

Business Units

August: Prelim. decisions on one-year

investment and allocation targets of

Business Units

Phase 1: Strategic Outlook for Business Units

January: Start identifying long-term investment opport. and present them to Division

CEO/CFO

Business Unit CEO

July: Negotiations over one-year investment

and allocation targets of Business UnitsDivision CEO

Business Unit CEO

Division CEO

Executive Board (Headquarters)

Phase 2: Annual Capital Allocation Plan for Business Units

Planning and Negotiations

Planning and Negotiations

Our data source: Annual Capital Allocation Plan

Panel B: Realization or Execution Stage

Involved Party:

Memo on Investment Proposal: Contains three documents: (1) Signing Form, (2) Investment

Description, (3) Output NPV Calculator

Memo on Investment Proposal: Contains three documents: (1) Signing Form, (2) Investment

Description, (3) Output NPV Calculator

Business Unit CEO

Division CEODivision CEO decides on

Proposal: Approval or Rejection

Executive Board (Headquarters)

Investment Proposal: Business Unit wants to execute an investment that has been

budgeted

NPV Calculator: Business Unit calculates NPV of

proposed investment using standardized excel

spreedshet

Headquarters provides inputs for NPV Calculator:

Interest Rates, Cost of Debt, Cost of Equity, Tax

Rates

Executive Board decides on Proposal: Approval or

Rejection

Execution of Investment (Order, Letter of Intent,

Signing of Contract, Payment, etc.)

Investment Authorization

Input

Our data source: Actual Capital Expenditures

If investment > Threshold Amount

If investment < Threshold Amount

If Approved

If Approved

Figure 2. Decision Process behind Allocation of Capital: Budgeting and Realization Stages. This figure presents the details of the capital allocation process of the conglomerate. Panel A shows the budgeting (i.e.,planning stage), and Panel B shows the realization (i.e., execution) stage. The budgeting consists of two different phases: the strategic outlook phase and the annual capital allocation phase. The figure shows how these two phases relate to each other, who the parties are, and when what kinds of decisions are made. A detailed verbal description of the capital allocation process is provided in Appendix A-3.

Planned versus Actual Capital Expenditures

0

0.01

0.02

0.03

0.04

0.05

0 0.01 0.02 0.03 0.04 0.05

Planned Capital Expenditures/Total Assets

Cap

ital E

xpen

ditu

res/

Tota

l Ass

ets

Figure 3. Planned versus Actual Capital Expenditures: Over-Budgeting by Business Units. This figure provides a scatter plot of planned capital expenditures (x-axis) and actual capital expenditures (y-axis) of the business units of the conglomerate. The figure uses annual averages of quarterly data from 01/2002 to 12/2006. The figure also contains at the 45-degree line to indicate whether planned capital expenditures are above (lower half of the figure) or below (upper half of the figure) actual expenditures. Planned capital expenditures are above (below) actual expenditures for 83% (17%) of the observations. This suggests that the business units systematically over budget their capital expenditures.

83% of obs.

17% of obs. 45° Line

Importance of Cash Windfalls Relative to Firm Level Variables

4% 6%24%

160%

82%

7% 13%

51%

194%

136%

10% 19%

84%

403%

126%

0%

50%

100%

150%

200%

250%

300%

350%

400%

450%

Cash Windfalls/Sales Cash Windfalls/Total Assets Cash Windfalls/Cash Flowfrom Operations

Cash Windfalls/Capital Exp. Cash Windfalls/EBIT

Perc

ent

Cash Windfall in 2003 Cash Windfalls in 2004 Cash Windfalls in 2005

Figure 4. Magnitude of Cash Windfalls. The figure documents the size of the cash windfalls relative to the conglomerate’s sales, total assets, cash flow from operations, capital expenditures, and EBIT. The conglomerate experienced eight cash windfalls resulting from headquarters’ sales of equity stakes in other companies. The equity sales occurred in the years 2003 (one sale), 2004 (three sales), and 2005 (four sales). The figure shows the aggregate value of the cash windfalls during each year relative to total aggregate conglomerate values of the same year.

Capital Expenditure Minus Planned Capital Expenditure in Cash Windfall and No Cash Windfall Quarters

-0.0060

-0.0024

-0.0049

-0.0024

-0.007

-0.006

-0.005

-0.004

-0.003

-0.002

-0.001

0.000

No Cash Windfall Quarters (14 Quarters) Cash Windfall Quarters (6 Quarters)

(Cap

ital E

xpen

ditu

re -

Plan

ned

Cap

ital

Expe

nditu

re)/T

otal

Ass

ets

Mean Median

Figure 5. Capital Expenditures in Cash-Windfall and No-Cash-Windfall Quarters. This figure illustrates actual minus planned capital expenditures, standardized by total assets, of the business units of the firm in cash-windfall and no-cash-windfall quarters. The table reports mean and median values, calculated over the respective business-unit quarters. Negative numbers suggest that actual investment is less than planned investment. The means (-0.0060 versus -0.0024) as well as the medians (-0.0049 versus -0.0024) are statistically different from each other at the 1% level.

-.0

4-.

02

0.0

2.0

4(C

api

tal E

xp. -

Pla

nned

Ca

pita

l Exp

.)/T

otal

Ass

ets

-.004 -.002 0 .002 .004Career at the Firm * log(1+Cash Windfall)

coef = 1.5768393, (robust) se = .81891347, t = 1.93

-.0

4-.

02

0.0

2.0

4(C

api

tal E

xp. -

Pla

nned

Ca

pita

l Exp

.)/T

otal

Ass

ets

-.004 -.002 0 .002 .004CEO Similarity* log(1+Cash Windfall)

coef = 1.021477, (robust) se = .50441265, t = 2.03

-.0

4-.

02

0.0

2.0

4(C

api

tal E

xp. -

Pla

nned

Ca

pita

l Exp

.)/T

otal

Ass

ets

-.004 -.002 0 .002 .004Power Index * log(1+Cash Windfall)

coef = 1.1736487, (robust) se = .64372367, t = 1.82

-.0

4-.

02

0.0

2.0

4(C

api

tal E

xp. -

Pla

nned

Ca

pita

l Exp

.)/T

otal

Ass

ets

-.003 -.002 -.001 0 .001 .002Networker * log(1+Cash Windfall)

coef = 3.0234449, (robust) se = 1.5631889, t = 1.93

-.0

4-.

02

0.0

2.0

4(C

api

tal E

xp. -

Pla

nned

Ca

pita

l Exp

.)/T

otal

Ass

ets

-.002 -.001 0 .001 .002Division CEO Connection * log(1+Cash Windfall)

coef = 2.2884953, (robust) se = 1.0788533, t = 2.12

-.0

4-.

02

0.0

2.0

4(C

api

tal E

xp. -

Pla

nned

Ca

pita

l Exp

.)/T

otal

Ass

ets

-.002 -.001 0 .001 .002Connected Networker * log(1+Cash Windfall)

coef = 3.6446177, (robust) se = 1.4372487, t = 2.54

Figure 6. Partial Scatter Plots. This figure presents partial scatter plots of the interaction term between a proxy for managerial power or connection and the cash-windfall variable. The dependent variable is the difference between actual and planned capital expenditures, scaled by total assets. Each regression uses the same controls and the same specifications as the regressions in Table VI.

Economic Effects of Power/Connection

2%

27%

10%

28%

2%

39%41%

36%

69%

50%

69%

-10%

0%

10%

20%

30%

40%

50%

60%

70%

80%

Career insideFirm

CEO Similarity Power Index Networker Division CEOConnection

ConnectedNetworker

Proxy for Managerial Power/Connection

(Cap

ital E

xp. -

Pla

nned

Cap

ital

Exp.

)/Tot

al A

sset

s

Low Pow er and Connection High Pow er and Connection

Figure 7. Economic Effect of Power and Connections. This figure illustrates the economic effects of power and connection. It reports the change in investment relative to planned investment (i.e., the change in actual minus planned investment, standardized by total assets) when our proxies for power or connection are low and high, respectively. A power or connection proxy is considered low (high) if the variable is at the 25th (75th) percentile. The changes in investment are calculated for a cash windfall of €938 million, which equals the average proceeds from cash windfalls in cash-windfall quarters. The figures are calculated based on the estimates in Table VI and correspond with those in Panel B of Table VI.

Appendix A-1: Definition of Variables

This table provides definitions of the variables used in the paper.

Variable Description 1. Financial Variables Number of Business Units Number of business units that operate under the umbrella of a division. Divisions

themselves have no operating activities. The data is from the flow chart of the firm and the organizational structure did not change over the sample period.

Capital Expenditures Investments in tangible and intangible assets (in EUR). We have both planned (budgeted) and actual data. The planned data is available on a quarterly basis, while the actual data is available on a monthly basis but added across the three months of a quarter to compare it with the planned data. The data is from the internal management accounting system of the firm.

Sales Proceeds from the sales of products or services to third parties and other divisions inside the firm (in EUR). We have both planned (budgeted) and actual data. The planned (budgeted) data is available on a quarterly basis, whereas the actual data is available on a monthly basis but added across the three months of a quarter to compare it with the planned data. The data is from the internal management accounting system of the firm.

Sales Growth Difference between this period’s sales and previous period’s sales divided by previous period’s sales. We have both planned (budgeted) and actual data. The data is from the internal management accounting system of the firm.

EBIT Earnings before interest payments and taxes (in EUR). We have both planned (budgeted) and actual data. The planned data is available on a quarterly basis, whereas the actual data is available on a monthly basis but added across the three months of a quarter to compare it with the planned data. The data is from the internal management accounting system of the firm.

EBIT Deviation from Plan Difference between the planned and actual value of EBIT (in EUR). The data is from the internal management accounting system of firm.

Total Assets Sum of the book values of fixed and current assets (in EUR). We have both planned (budgeted) and actual data. The planned data is available on a quarterly basis, whereas the actual data is available on a monthly basis but averaged across the three months of a quarter to compare it with the planned data. The data is from the internal management accounting system of the firm.

Cash Flow from Operations Net income, plus depreciation and amortization, and plus (minus) decreases (increases) in working capital (in EUR). We have both planned (budgeted) and actual data. The planned data is available on a quarterly basis, whereas the actual data is available on a monthly basis but added across the three months of a quarter to compare it with the planned data. The data is from the internal management accounting system of the firm.

Personnel Number of employees, calculated as full-time equivalents. We have both planned (budgeted) and actual data. The planned data is available on a quarterly basis, whereas the actual data is available on a monthly basis but averaged across the three months of a quarter to compare it with the planned data. The data is from the internal management accounting system of the firm.

R&D Expenditures Expenses related to research and development activities (in EUR). We have both planned (budgeted) and actual data. The planned data is available on a quarterly basis, while the actual data is available on a monthly basis but added across the three months of a quarter to compare it with the planned data. The data is from the internal management accounting system of the firm.

Marketing Expenditures Expenses related to the marketing and selling of products and services (in EUR). We have both planned (budgeted) and actual data. The planned data is available on a quarterly basis, while the actual data is available on a monthly basis but added across the three months of a quarter to compare it with the planned data. The data is from the internal management accounting system of the firm.

Imputed Tobin’s Q The median value of Tobin’s Q for all firms from the European Union with the same three-digit SIC code. The variable is constructed based on data from Compustat and Compustat Global.

Total Debt Total interest-bearing debt (in EUR or USD). The data is from Compustat and Compustat Global.

Cash Holdings Cash and marketable securities (in EUR or USD). The data is from Compustat and Compustat Global.

Dividends Dividends paid to shareholders (in EUR or USD). The data is from Compustat and Compustat Global.

Business Unit Equal to one if an observation is from one of the twenty business units of the

conglomerate, and zero otherwise. Fourth Quarter Dummy Equal to one if an observation is from the fourth quarter of the calendar year, and zero

otherwise. The variable is included to account for the seasonality in the business of the firm.

2. Cash Windfall Variable Cash Windfall Cash inflow at headquarters that results from the selling of equity stakes in other firms (in

million EUR). Headquarters had equity stakes in other firms that were sold and resulted in eight large cash windfalls during the sample period. The variable is self-constructed based on annual reports from the firm, press statements, and a web search.

3. Managerial Power and Connection Variables Career at the Firm This index measures the career a manager has made at the conglomerate and a manager’s

general social network at the workplace. It is formed by averaging the following three variables: (1) Number of months during which a manager had an overlap inside a unit of the conglomerate with a person, which later became an executive board member. (This can be the case, for example, if a manager worked at the same time in the same business unit of the firm as the CFO); (2) Number of years a manager has been working for the firm; (3) Number of years a manager has been working in a powerful position inside the firm (e.g., division CEO, business-unit CEO, or head of marketing). All three variables are normalized before entering the index to fall between zero and one, using the respective minimum and maximum values. All variables entering the index are self-constructed based on data from BoardEx, ZoomInfo, social networking web pages (LinkedIn, Facebook, MySpace), annual reports, and systematic web and newspaper searches. This data is complemented with data from a survey among the managers, which we developed and ran ourselves. Details on the index construction and the variables it contains are reported in Appendix A-2.

CEO Similarity This index measures the similarity of the personal profile of a manager with that of the CEO. It is formed by averaging the following four dummy variables: (1) Dummy, which is equal to one if a manager speaks the native language of the CEO; (2) Dummy, which is equal to one if a manager lives in the country in which the CEO lives; (3) Dummy, which is equal to one if a manager went to the same university as the CEO; (4) Dummy, which is equal to one if a manager graduated in the same academic discipline (engineering) as the CEO. All variables entering the index are self-constructed based on data from BoardEx, ZoomInfo, social networking web pages (LinkedIn, Facebook, MySpace), annual reports, and systematic web and newspaper searches. This data is complemented with data from a survey among the managers, which we developed and ran ourselves. Details on the index construction and the variables it contains are reported in Appendix A-2.

Power Index This index is formed by averaging the two indexes: (1) Career at the Firm; and (2) CEO Similarity.

Networker This index measures the extent to which a manager undertakes networking activities. The index is formed by averaging the following five dummy variables: (1) Dummy, which is equal to one if a manager is a member in a fraternity; (2) Dummy, which is equal to one if a manager is a member of a social club inside the firm; (3) Dummy, which is equal to one if a manager regularly stops by headquarters to say “Hello”; (4) Dummy, which is equal to one if a manager regularly accepts highly visible work assignments; (5) Dummy, which is equal to one if a manager participated regularly in highly visible task forces or committees. All variables entering this index are self-constructed based on a survey, which we developed and ran ourselves among the managers. The survey questions underlying items (3)-(5) are based on Forret and Dougherty (2001, 2004). Details on the index construction and the variables it contains are reported in Appendix A-2.

Division CEO Connection This index measures how well a manager is connected to a division CEO. The index is formed by averaging the following variables: (1) Dummy, which is equal to one if a manager has a division CEO as a personal mentor; (2) Dummy, which is equal to one if a manager goes at least occasionally for lunch with the Division CEO; (3) Dummy, which is equal to one if a manager meets the division CEO in person at least every two weeks. All variable entering this index are self-constructed based on a survey, which we developed and ran ourselves among the managers. The survey question underlying item (1) is based on Podolny and Baron (1997), and the survey question underlying item (2) is based on Forret and Dougherty (2001, 2004). Details on the index construction and the variables it contains are reported in Appendix A-2.

Connected Networker This index is formed by averaging the two indexes: (1) Networker; and (2) Division CEO Connection.

4. Managerial Ability Variables External Director Equal to one if a manager holds a non-executive director position outside the

conglomerate in addition to his management position inside the firm during the period of our analysis (i.e., 2002-2006), and zero otherwise. The variable is self-constructed based on data from BoardEx, ZoomInfo, social networking web pages (LinkedIn, Facebook, MySpace), annual reports, and systematic web and newspaper searches. This data has been complemented with data from a survey among the managers, which we developed and ran ourselves.

Elite University Equal to one if a manager has a degree from an elite university, and zero otherwise. For our analysis, elite universities include the Ivy League schools in the US as well as the five leading business schools in Europe according to the Financial Times Business School Ranking. The variable is self-constructed based on data from BoardEx, ZoomInfo, social networking web pages (LinkedIn, Facebook, MySpace), annual reports, and systematic web and newspaper searches. This data has been complemented with data from a survey among the managers, which we developed and ran ourselves.

Appendix A-2: Details on Components of Power and Connection Indices

This table provides details on the components of the power and connection indices used in the paper. It contains details on the coding of the variables as well as summary statistics. 1. Career at the Firm (1) Number of months during which a manager coincided at a unit of the conglomerate with a person who later became an executive board member.

Mean Median Std. dev. Obs. 62 24 92 39

(2) Number of years a manager has been working for the firm.

Mean Median Std. dev. Obs. 12.1 5.2 11.0 39

(3) Number of years a manager has been working in a powerful position inside the firm.

Mean Median Std. dev. Obs. 5.5 3.5 5.7 37

2. CEO Similarity (1) Dummy, which is equal to one if a manager speaks the native language of the CEO (=Yes).

Yes No Obs.

52.5% 47.5% 40 (2) Dummy, which is equal to one if a manager lives in the country in which the CEO lives (=Yes).

Yes No Obs. 70% 30% 40

(3) Dummy, which is equal to one if a manager went to the same university as the CEO (=Yes).

Yes No Obs. 11% 89% 37

(4) Dummy, which is equal to one if a manager graduated in the same academic discipline (engineering) as the CEO (=Yes).

Yes No Obs. 53% 47% 36

3. Power Index This index builds on the previous two indices. 4. Networker (1) Are you a member of a fraternity? _ Yes _ No

Yes No Obs.

30% 70% 20 (2) During your time as business-unit CEO, were you a member of one of the following clubs at (COMPANY)? Club 1: _ Yes _ No; Club 2: _ Yes _ No; Club 3: _ Yes _ No

Yes No Obs.

21% 79% 19 We code this dummy as equal to one if a manager is a member of at least one club. (3) During your last year as business-unit CEO, how often did you conduct the following activity to increase your visibility within (COMPANY) (please mark)? Stopping by division headquarters to say “hello”?

Never Seldom, only

once or twice Occasionally, several

times a year Moderately often,

every few week Often, almost every week

Very often, almost every

day

Obs.

11% 24% 12% 23% 18% 12% 17 We create a dummy for this survey question, which is equal to one if a manager regularly stops by headquarters to say “hello”. This is the case if the answer is “moderately often, every few weeks”, or more frequently. (4) During your last year as business-unit CEO, how often did you conduct the following activity to increase your visibility within (COMPANY) (please mark)? Accepting new, highly visible work assignments?

Never Seldom, only once or twice

Occasionally, several times a year

Moderately often, every few weeks

Often, almost every week

Very often, almost every

day

Obs.

18% 18% 35% 18% 6% 6% 17

We create a dummy for this survey question, which is equal to one if a manager regularly accepts new visible work assignments. This is the case if the answer is “Occasionally, several times a year”, or more frequently. (5) During your last year as business-unit CEO, how often did you conduct the following activity to increase your visibility within (COMPANY) (please mark)? Being on highly visible task forces or committees at work?

Never Seldom, only once or twice

Occasionally, several times a year

Moderately often, every few weeks

Often, almost every week

Very often, almost every

day

Obs.

17% 17% 44% 11% 6% 6% 18 We create a dummy for this survey question, which is equal to one if a manager regularly sits on highly visible task forces or committees at work. This is the case if the answer is “Occasionally, several times a year”, or more frequently. 5. Division CEO Connection (1) During your time as business-unit CEO/CFO, were there any division CEOs you regarded as a guide/mentor—that is, someone who has taken a strong interest in your professional development over the last year at (COMPANY) /in the past by providing you with opportunities and/or access to facilitate your career advancement?

Yes No Obs. 50% 50% 18

(2) During your last year as business-unit CEO, how often did you conduct the following activities to increase your visibility within (COMPANY) (please mark)? Going to lunch with your division CEO?

Never Seldom, only once or twice

Occasionally, several times a year

Moderately often, every few weeks

Often, almost every week

Very often, almost every

day

Obs.

23% 23% 29% 12% 12% 0% 17 We create a dummy for this survey question, which is equal to one if a manager regularly goes for lunch with his division CEO “occasionally, several times a year”, or more frequently. (3) During your time as business-unit CEO/CFO, how often did you meet the CEO of your division in person?

Every day Twice a week Once a week Every two weeks Once a month Not during the last month

Obs.

0% 26% 11% 16% 32% 16% 19 We create a dummy for this survey question, which is equal to one if a manager meets the division CEO “every two weeks”, or more frequently. 6. Connected Networker This index builds on the previous two indices.

Appendix A-3: Detailed Description of the Capital Allocation Process

To understand whether and where the intra-firm bargaining power of managers plays a

role for capital allocations, we have access to information describing the capital allocation process

of the firm. We can use documents on (i) the institutional details of the allocation process, provided,

for example, in the budgeting manual of the firm, (ii) the time-line of budgeting meetings, (iii) the

managers who participate in these meetings, (iv) the approval procedures for investments, and (v) a

net present value calculator for project evaluation. The internal capital allocation process itself is

reported consists of two general stages: a budgeting or planning stage and a realization or execution

stage.

A. Budgeting or Planning Stage

In the budgeting stage, the firm transforms the general corporate strategy into capital

allocations (budgets) using a two-phase process, which is anchored on the elaboration of two key

documents for each business unit: (1) the strategic outlook, and (2) the annual capital allocation

plan. The firm has a very standardized two-phase process for transforming this general corporate

strategy into concrete capital allocations and investment decisions.

Phase 1: Strategic Outlook for Business Units

The objective of the first phase is to develop a three-year strategic outlook (“guide”) for all

business units of the firm. An important aspect of the strategic outlook is to have discussions

between division and business-unit CEOs and the executive board on long-run strategies of the

divisions and their business units.

Therefore, in January the business-unit CEOs start identifying long-term growth and

investment opportunities of their business units and the lines of business operating within the

business units. These opportunities refer to projects with horizons over several years and are

supposed to be consistent with the general strategy of the firm, as outlined by the executive board.

The developed business plans are then presented by the business-unit CEOs to the respective

division CEOs. In February, negotiations in special budget meetings take place between the

business-unit CEOs and the corresponding division CEOs over these identified opportunities, with

preliminary decisions on the implied business plans being made by the division CEOs in March.

Each division CEO and his business-unit CEOs then present the three-year business plans of their

units, using standardized forms (so-called budget forms), to the executive board and negotiate over

revisions and adjustments.

The negotiations are coordinated, scheduled, and structured by corporate control, a

centralized support function at headquarters, which also provides sample forms and records as well

as a budget manual to the divisions and their business units. Corporate control runs additional

checks and analyses of the presented plans and attempts to ensure that the process is completed

according to the budget time-line. Corporate control also issues guidelines to the divisions to ensure

that the strategies of the divisions are consistent with the overall strategy of the firm and to

standardize the documents, financials and presentations that are provided to the executive board.

Following the negotiations among the executive board and the unit managers, a final decision on

the figures in the strategic outlook is being made in a meeting of the executive board in April.

The strategic outlook includes general targets for planned investments and required resource

allocations for all units. It thereby serves as the foundation for the internal capital allocation process.

As described, the process is highly institutionalized and structured, with decisions being made for

the three-year period starting in January of the subsequent year.

Phase 2: Annual Capital Allocation Plan for Business Units

Just after the finalization of the strategic outlook, a concrete annual capital allocation plan

(or capital budget) is developed for the coming year. The annual capital allocation plan is a one-

year investment plan containing detailed resource allocations for all business units.

The link between the strategic outlook and the allocation plan is the breaking down of the

business plans for the setting of very concrete investment targets. The annual capital allocation plan

usually starts with the projection of revenues and associated costs for the respective business units,

usually build on estimates per line of business and sometimes even per product. Based on the

estimates, the units derive the investment and capital allocation targets needed to fulfill these

projections. These targets comprise targets for capital expenditures, marketing expenditures, R&D

expenditures, and, at headquarters, expenditures for M&A. The annual allocation plan therefore

translates into a complete set of pro forma balance sheets, income statements, and cash flow

statements for each business unit, which are subsequently aggregated and consolidated at division

as well as firm level.

Discussions about the annual allocation plan start in June with the division CEOs, based on

the strategic outlook, preparing investment and allocation targets for the coming year. The

business-unit and division CEOs then jointly negotiate in July over these allocation plans and make

revisions and adjustments. Preliminary decisions on the budgets are made in August. As in the

strategic outlook, the division and business-unit CEOs then present the capital allocation plan for

their divisions to the executive board using standardized forms and a second round of in-depth joint

negotiations and revisions takes place. These negotiations take place in September and October. As

in the strategic outlook, the negotiations are coordinated, scheduled, and structured by corporate

control. Finally, in October or November, the executive board decides on the investment and

allocation targets for the coming year. The plan is then presented by the executive board to the

supervisory board for final approval, which is usually a formality.

The divisions and business units then receive the approved plan and prepare the

implementations for the coming year. As illustrated in Appendix A-7, our data source is this annual

capital allocation plan. We have data from five annual capital allocation plans, from 2002 to 2006,

each containing quarterly plans and budgets for all 20 business units of the firm. Once the annual

capital allocation plan is approved, it also serves as a foundation for headquarters’ financing and

fiscal planning for the year ahead.

As suggested by the above descriptions, the budgeting process and the resulting capital

allocation of the firm tries to combine top-down and bottom-up elements to enable collaboration

and information aggregation between lower and top management.

B. Realization or Execution Stage

It is during the realization or execution stage that concrete investments are made and the

corresponding cash outflows are generated. Decisions here are made much more quickly than they

are during budgeting. All investments (buying or replacing small machines or IT, for example)

below the investment thresholds can be made, quickly and easily, with simple procedures at the

discretion of the business units.

Although the annual capital allocation plan is the basis for investments by the business units,

the firm requires additional approvals for the execution of any project in the plan that involves

investments or divestments in tangible or intangible fixed assets that exceed a specific threshold.

The thresholds depend on the divisions and on their capital intensity.

If the proposed investment exceeds the threshold, the business unit must prepare, for

approval by the division CEO and the executive board, an investment memo that contains (1) a

document to be signed by the executive board, (2) a document describing the investment, and (3) a

net present value calculation. The project description should explain the reason for and the risks the

investment and discuss of the alternatives. Each business unit must show that the investment will

generate as a minimum requirement a positive NPV. For this reason, the company has constructed

an NPV calculator, a spreadsheet provided by headquarters that all units must use to evaluate

projects.

As the discount rate for the NPV calculation, the firm uses a project-specific weighted

average cost of capital (WACC), provided by headquarters. The input on interest rates, cost of debt

and country-risk premia used for the calculation of the WACC are provided by the corporate

treasury, a department at the headquarters. Tax rates are provided by the corporate tax department,

also at headquarters.

The NPV of a project is not the only relevant approval criteria and when it comes to

deciding which positive NPV projects are being executed, other things such as strategic fit, the cash

flow structure, past performance or reputation of the business-unit CEOs also play a role.

Taking into account all factors, the executive board makes a final decision on the

investment approval and subsequently informs the business unit of its final decision by automated

email. After that, the business units can place an order, assign a letter of intent or contract or make

a payment. Headquarters then provides the corresponding actual allocation of funds so that the

investment can be executed.

Two years after the approval of a project by the executive board, the firm requires the filing

of a post-investment report that documents the extent to which was consistent with the figures in

the NPV calculator, and where and why deviations occurred.

The company applies the OECD principles on transfer prices for all within-firm sales. In

other words, transfers within the firm are priced using the arm’s-length principle, that is, transfer

prices should be the same as if the two divisions or business units involved were two independent

firms not part of the same conglomerate. Any significant reallocation of capital from one unit to

another through transfer prices is therefore unlikely.

Appendix A-4: Descriptive Statistics This table presents descriptive statistics of the variables used in the paper. The variables are calculated based on quarterly data from 01/2002 to 12/2006. Panel A provides descriptive statistics of financial variables of the business units. The statistics are based on data from the internal management accounting system of the conglomerate and calculated across all twenty business units of the conglomerate. Panel B reports statistics of the cash-windfall variable. The statistics are calculated and reported across the six cash - quarters. Panel C reports variables proxying for the power and connection of the business-unit CEOs. Panel D provides variables that proxy for ability of business-unit CEOs. Detailed definitions of each variable are provided in Appendix A-1. Correlations are reported in Appendix A-6. Panel A: Financial Variables of Business Units

Obs. Mean Median Std. dev. 25% 75%

Planned Sales (millions) 391 369 227 386 121 416

Planned EBIT/Total Assets 391 0.043 0.038 0.065 0.006 0.069

Planned Sales Growth 371 0.058 0.057 0.268 -0.016 0.163

Planned Cash Flow from Operations/Total Assets 215 0.052 0.046 0.102 0.009 0.099

Planned Capital Expenditures (millions) 359 12.9 4.9 25.5 3.3 11.0

Planned Capital Expenditures/Total Assets 359 0.016 0.012 0.013 0.006 0.023

Planned R&D Expenditures/Total Assets 211 0.040 0.030 0.041 0.021 0.044

Planned Marketing Expenditures/Total Assets 215 0.101 0.077 0.087 0.052 0.134

Planned Personnel (# of employees) 215 6780 4113 7976 2725 7956

Sales (millions) 391 345 219 370 108 392

EBIT/Total Assets 391 0.032 0.034 0.096 -0.006 0.068

Sales Growth 370 0.041 0.026 0.270 -0.085 0.153

Cash Flow from Operations/Total Assets 391 0.051 0.043 0.118 0.001 0.099

Capital Expenditures (millions) 391 9.6 3.5 21.9 1.8 8.4

Capital Expenditures/Total Assets 391 0.011 0.007 0.013 0.004 0.017

R&D Expenditures/Total Assets 391 0.035 0.029 0.029 0.018 0.047

Marketing Expenditures/Total Assets 391 0.098 0.080 0.066 0.049 0.148

Personnel (# of employees) 391 6537 3929 8028 1982 7657

Imputed Tobin's Q 391 1.690 1.799 0.396 1.324 1.961

(Capital Expenditures - Planned Capital Expenditures)/Total Assets 359 -0.005 -0.004 0.011 -0.010 0.001

EBIT Deviation from Plan/Total Assets 311 0.000 0.001 0.021 -0.009 0.012

Panel B: Cash-Windfall Variable

Obs. across Cash-Windfall Quarters (i.e. if cash windfalls>0) Obs. Mean Median Std. dev. 25% 75%

Cash Windfall (millions) 132 938 903 258 672 1265

Panel C: Managerial Power Variables

Obs. Mean Median Std. dev. 25% 75%

Career at the Firm 321 0.252 0.163 0.234 0.076 0.381

CEO Similarity 346 0.486 0.500 0.213 0.250 0.500

Power Index 313 0.377 0.311 0.198 0.264 0.485

Networker 174 0.468 0.400 0.202 0.400 0.600

Division CEO Connection 166 0.552 0.667 0.274 0.333 0.667

Connected Networker 166 0.497 0.533 0.198 0.367 0.633

Panel D: Managerial Ability Variables Obs. Mean Median Std. dev. 25% 75% External Director 387 0.32 0.00 0.47 0.00 1.00 Elite University 354 0.35 0.00 0.48 0.00 1.00

Appendix A-5: Description of Cash Windfalls

This table presents a summarized description of eight cash windfalls, which occurred at the conglomerate’s headquarters. During the sample period of 2002 to 2006, the headquarters of the conglomerate sold eight equity stakes in other firms and generated cash inflows of more than €500 million for each sale. The cash windfalls occurred in six different quarters—Q4 2003, Q3 2004, Q4 2004 (two windfalls), Q2 2005, Q3 2005 (two windfalls), and Q4 2005. Panel A presents whether the stakes were related to the activities of the business units, the stated reasons for the sale, the procedure by which the stake was sold, and the size of the pre- and post-sale stake (in % of the shares of the firm whose equity was sold). It also contains information on how long the firm has held the equity stakes. Panel B presents information on the use to which the windfalls were put. The statistics in Panel A are constructed based on information from company news reports as well as newspaper articles. The figures in Panel B are calculated based on the cash flow statements of the firm in the quarters in which cash windfalls occurred. We assume that an additional cash from the cash windfalls is, on aggregate, used in the same way as an additional cash from other operating or financing sources in a cash-windfall quarter. Panel A: Description of Cash Windfalls Unrelated to Activities of Business Units Yes 5 (# of Equity Stakes) No 3 Stated Reason for Equity Sale # Non-strategic asset 6 (multiple reasons possible) # Exploit high market value 3 # Reduce exposure to cyclical industry 2 Procedure of Sale of Equity Stake # Initial Public Offering (IPO) 1 # Seasoned Equity Offering (SEO) 4 # OTC Transaction (OTC) 3 Pre-Sale Equity Stake Mean 32% Median 29% After-Sale Equity Stake Mean 16% Median 16% Holding Period of Equity Stake Mean 12 (years) Median 11 Panel B: Usage of Cash Windfalls Capital Expenditures 19% Increase in Working Capital 1% Acquisitions 13% Repayment of Debt 21% Dividends 9% Share Repurchases 11% Tax 6% Increase in Cash Holdings 20%

Appendix A-6: Correlations of Main Variables

This table provides pairwise correlations of the main variables in the data set. The variables are defined in Appendix A-1 and descriptive statistics reported in Appendix A-4. Correlations are calculated based on observations at the business-unit-quarter. (1) (2) (3) (4) (5) (6) (7) (8) (9) (10) (11) (12) (13) (14) (15) (16) Sales (1) 1.00 EBIT/Total Assets (2) 0.33 1.00 Sales Growth (3) 0.10 0.24 1.00 Cash Flow from Operations/Total Assets (4) 0.33 0.77 0.45 1.00 Capital Expenditures/Total Assets (5) 0.04 0.34 0.14 0.36 1.00 R&D Expenditures/Total Assets (6) -0.32 -0.54 0.03 -0.35 -0.06 1.00 Marketing Expenditures/Total Assets (7) -0.17 -0.06 0.22 0.14 0.27 0.71 1.00 Imputed Tobin's Q (8) -0.12 -0.27 0.02 -0.24 -0.31 0.27 0.11 1.00 Career at the Firm (9) 0.47 0.18 -0.04 0.17 0.28 -0.22 0.01 -0.39 1.00 CEO Similarity (10) 0.19 0.05 -0.03 0.07 0.29 -0.06 0.02 -0.63 0.73 1.00 Power Index (11) 0.36 0.13 -0.04 0.13 0.31 -0.16 0.02 -0.54 0.94 0.92 1.00 Networker (12) -0.20 0.26 0.04 0.20 0.34 -0.15 0.08 -0.20 0.22 0.07 0.16 1.00 Division CEO Connection (13) 0.09 0.27 0.03 0.25 0.36 -0.20 0.09 -0.49 0.42 0.38 0.43 0.50 1.00 Connected Networker (14) -0.04 0.30 0.04 0.26 0.41 -0.20 0.10 -0.42 0.39 0.29 0.37 0.81 0.91 1.00 External Director (15) 0.36 0.20 -0.07 0.08 -0.19 -0.48 -0.59 -0.13 0.17 0.07 0.13 -0.44 -0.02 -0.22 1.00 Elite University (16) 0.40 0.18 -0.05 0.07 -0.17 -0.50 -0.62 -0.12 0.16 0.02 0.10 -0.35 0.02 -0.15 0.95 1.00

Appendix A-7: Capital Budgeting Process over Time and Available Data

This figure presents a time-line of the processes behind the allocation of capital. It also reports the periods for which we have data available. A detailed verbal description of the allocation process is provided in Appendix A-3.

Meetings

Meetings

Meetings

Meetings

Meetings

Meetings

Q1 Q2 Q3 Q1 Q2 Q3 Q1 Q2 Q3 Q1 Q2 Q3 Q1 Q2 Q3 Q1 Q2 Q3

Availability of Data:

Actual Data

Plan Data

means: "Decision over plan is made"

Capital Budgeting Process at ConglomerateBudgeting Process for:

Years 2002-2004Strategic Outlook

Meetings Annual Capital Allocation Plan

Years 2003-2005Strategic Outlook

Meetings Annual Capital Allocation Plan

Years 2004-2006Strategic Outlook

Meetings Annual Capital Allocation Plan

Years 2005-2007Strategic Outlook

Meetings Annual Capital Allocation Plan

Years 2006-2008Strategic Outlook

Meetings Annual Capital Allocation Plan

Years 2007-2009Meetings

Q4 Q4 Q4 Q4 Q4 Q42005 2006

Annual Capital Allocation Plan Data (2002-2006)

2001 2002 2003 2004

Actual Capital Allocation Data (2002-2006)

D.D.

D.D.

D.D.

D.D.

D.D.

D.

D.D.

Appendix A-8: Performance, Capital Allocation, and Managerial Ability This table presents OLS regressions for the twenty business units of the conglomerate. The dependent variable (shown in the first row) in regressions (1) and (2) is a business unit’s quarterly sales growth and in regressions (3) and (4) a business unit’s quarterly planned capital expenditures over its total assets over the period from 01/2002 to 12/2006. Regressions (1) and (3) include Elite University as controls. Regressions (2) and (4) include External Director as controls. All regressions also control for: Personnel/Total Assets; Cash Flow from Operations/Total Assets; Capital Expenditures/Total Assets (lag); Fourth-Quarter Dummy; R&D Expenditures/Total Assets; Marketing Expenditures/Total Assets; Sales; Planned EBIT/Total Assets (lag); Planned Personnel/Total Assets; Planned Sales Growth; EBIT Deviations from Plan/Total Assets (lag); and a constant (not shown). The number of business-unit quarters and business units in each regression are reported at the bottom of the table. Robust standard errors, clustered at business-unit level, are shown in brackets. All dependent variables and controls are Winsorized at 1%. Detailed definitions of each variable are provided in Appendix A-1. *** indicates significance at 1%, ** indicates significance at 5%, and * indicates significance at 10%.

Sales Growth Planned Capital

Expenditure/Total Assets

[1] [2] [3] [4]

Elite University 0.0715*** 0.0061**

[0.0208] [0.0022]

External Director -0.0064 -0.0014

[0.0278] [0.0018]

Personnel/Total Assets -0.0097 -0.0064

[0.0072] [0.0077]

Cash Flow from Operations/Total Assets 0.4055** 0.3503**

[0.1517] [0.1622]

Capital Expenditures/Total Assets (lag) -4.1002** -4.5186***

[1.4565] [1.4077]

Fourth-Quarter Dummy 0.1967*** 0.2079*** -0.0023** -0.0021**

[0.0493] [0.0488] [0.0010] [0.0010]

R&D Expenditures/Total Assets -1.9871 -2.1000

[1.5547] [1.7535]

Marketing Expenditures/Total Assets 2.6834*** 2.6106***

[0.7516] [0.8268]

Sales 0.0007*** 0.0007***

[0.0002] [0.0002]

Planned EBIT/Total Assets (lag) 0.0290** 0.0321**

[0.0110] [0.0116]

Planned Personnel/Total Assets 0.0005** 0.0005**

[0.0002] [0.0002]

Planned Sales Growth 0.0008 -0.0017

[0.0043] [0.0053]

EBIT Deviation from Plan/Total Assets (lag) 0.0120 0.0017

[0.0291] [0.0312]

Business-Unit Fixed Effects Yes Yes Yes Yes

Year Fixed Effects Yes Yes Yes Yes

Observations 353 386 209 211 Number of Business Units 20 20 18 18

Adjusted R-squared 0.496 0.484 0.094 0.068

Appendix A-9: Effects of Managerial Power and Connection for R&D and Marketing Expenditures

This table shows OLS regressions for the twenty business units of the conglomerate. The dependent variable (shown in the first row) in Panel A (Panel B) is the quarterly difference between a business unit's actual and planned R&D (Marketing) expenditures over its total assets over the period from 01/2002 to 12/2006. We report seven regressions in each panel. The first regression of each panel does not include any managerial power index as a control. The following six regressions of each panel successively control for the managerial power indices indicated in the second row: Career at the Firm; CEO Similarity; Power Index; Networker; Division CEO Connection; and Connected Networker. The last six regressions of each panel also control for the interaction of each of the managerial power indices and Ln(1+Cash Windfall). The coefficient of the interaction term and its standard error are multiplied by 1,000. All regressions of each panel also control for: Ln(1 + Cash Windfall); Sales Growth; EBIT/Total Assets; EBIT Deviation from Plan/Total Assets (lag); Imputed Tobin's Q (lag); Fourth-Quarter Dummy; and a constant (not shown). The number of business-unit quarters and business units in each regression are reported at the bottom of the table. Robust standard errors, clustered at business-unit level, are shown in brackets. All dependent variables and the controls are Winsorized at 1%. Detailed definitions of each variable are provided in Appendix A-1. *** indicates significance at 1%, ** indicates significance at 5%, and * indicates significance at 10%. Panel A: R&D Expenditures Dependent Variable: (R&D Expenditures - Planned R&D Expenditures)/Total Assets

Managerial Power Variable: None Career at the Firm

CEO Similarity

Power Index Networker

Division CEO

Connection Connected Networker

[1] [2] [3] [4] [5] [6] [7]

Managerial Power * Ln(1+Cash Windfall) 0.7640 0.9182 0.7344 0.9423 0.4234 1.2261

[0.8484] [0.9123] [0.9253] [1.4646] [1.0386] [1.3923]

Managerial Power -0.0184 -0.0076 -0.0166 0.0104 0.0050 0.0113

[0.0184] [0.0098] [0.0108] [0.0131] [0.0121] [0.0324]

Ln(1+Cash Windfall) -0.0002 -0.0003 -0.0005 -0.0003 -0.0002 0.0001 -0.0003

[0.0002] [0.0002] [0.0005] [0.0003] [0.0006] [0.0006] [0.0007]

Sales Growth 0.0135* 0.0127* 0.0130* 0.0130 0.0145 0.0143

[0.0066] [0.0063] [0.0063] [0.0080] [0.0082] [0.0083]

EBIT/Total Assets -0.0721** -0.0681** -0.0719** -0.0901** -0.0909** -0.0905**

[0.0264] [0.0285] [0.0269] [0.0318] [0.0328] [0.0324]

EBIT Deviation from Plan/Total Assets (lag) -0.0037 -0.0206 -0.0132 -0.0016 0.0135 0.0064

[0.0576] [0.0426] [0.0563] [0.0480] [0.0506] [0.0509]

Imputed Tobin's Q (lag) -0.0015 0.0036 0.0020 -0.0761 -0.0795 -0.0784

[0.0193] [0.0147] [0.0164] [0.1229] [0.1319] [0.1310]

Fourth-Quarter Dummy -0.0012 -0.0006 -0.0013 0.0002 -0.0003 -0.0004

[0.0018] [0.0016] [0.0017] [0.0019] [0.0020] [0.0021]

Business-Unit Fixed Effects Yes Yes Yes Yes Yes Yes Yes

Year Fixed Effects Yes Yes Yes Yes Yes Yes Yes

Observations 211 181 205 181 103 95 95

Number of Business Units 18 16 18 16 11 10 10

Adjusted R-squared -0.002 0.046 0.058 0.045 0.078 0.072 0.075

Appendix A-9 (continued)

Panel B: Marketing Expenditures Dependent Variable: (Marketing Expenditures - Planned Marketing Expenditures)/Total Assets

Managerial Power Variable: None Career at the Firm

CEO Similarity

Power Index Networker

Division CEO

Connection Connected Networker

[1] [2] [3] [4] [5] [6] [7]

Managerial Power * log(1+Cash Windfall) 2.2399 5.1189 4.8517 6.5403 -0.4310 2.6719

[2.6102] [4.4259] [4.8850] [5.5746] [3.9678] [5.3792]

Managerial Power -0.0506 0.0125 -0.0022 0.0296 -0.0473 -0.1152

[0.0737] [0.0099] [0.0241] [0.0241] [0.0473] [0.1048]

log(1+Cash Windfall) -0.0007 -0.0003 -0.0023 -0.0016 -0.0022 0.0006 -0.0010

[0.0008] [0.0008] [0.0022] [0.0019] [0.0023] [0.0016] [0.0022]

Sales Growth 0.0126* 0.0121* 0.0118* 0.0194 0.0259 0.0274*

[0.0065] [0.0058] [0.0064] [0.0139] [0.0142] [0.0145]

EBIT/Total Assets 0.0557 0.0470 0.0525 0.0404 0.0230 0.0215

[0.0435] [0.0326] [0.0403] [0.0541] [0.0533] [0.0542]

EBIT Deviation from Plan/Total Assets (lag) -0.2373 -0.2276 -0.2574 -0.2751 -0.1972 -0.1277

[0.1658] [0.1428] [0.1821] [0.2120] [0.1880] [0.1294]

Imputed Tobin's Q (lag) -0.0418 -0.0445 -0.0359 -0.3575 -0.3942 -0.4078

[0.0444] [0.0334] [0.0283] [0.2894] [0.3102] [0.3192]

Fourth-Quarter Dummy -0.0033 -0.0017 -0.0027 -0.0088 -0.0091 -0.0084

[0.0041] [0.0029] [0.0036] [0.0061] [0.0061] [0.0066]

Business-Unit Fixed Effects Yes Yes Yes Yes Yes Yes Yes

Year Fixed Effects Yes Yes Yes Yes Yes Yes Yes

Observations 359 246 270 242 141 133 133

Number of Business Units 20 18 20 18 13 12 12

Adjusted R-squared 0.034 0.100 0.106 0.089 0.076 0.064 0.075